CN114416912A - Event description text prediction method, device, device and storage medium - Google Patents

Abstract

The application provides a method, a device, equipment and a computer readable storage medium for predicting description texts of events, wherein the method comprises the following steps: acquiring an event description text of a target event from a source terminal; performing semantic analysis on the event description text to obtain key verbs and key nouns; determining the matching degree of the key verb and a first node of the nodes in the event relation tree based on the event relation tree; determining a target node in the event relation tree according to the first node matching degree to obtain a target first text and a target second text; determining a first degree of association of the nouns in the target first text with the nouns in the target second text; determining a target noun corresponding to the key noun according to the first association degree and the key noun; and carrying out noun replacement processing on the target second text according to the target noun to obtain a target description text, and sending the target description text to the target terminal. The application also relates to blockchain technology, and the target description text can be stored in the blockchain.

Description

Event description text prediction method, device, device and storage medium

technical field

The present application relates to the technical field of text prediction, and in particular, to a method, apparatus, device, and computer-readable storage medium for event description text prediction.

Background technique

With the frequent occurrence of emergencies in recent years, the whole society is paying more and more attention to emergencies. Without adequate preparation, emergencies are often caught off guard, and when life safety is involved, time is life. Therefore, only quick and accurate responses can escort life safety. The relevant experience accumulated in the handling of more and more emergencies can be used as a preparatory measure to deal with future emergencies, and the development of events can provide great help for the prediction of emergencies.

Most of the current work uses probabilistic and statistical correlation methods based on word co-occurrence relationships to describe the logical laws and behavior patterns between events. For example, given the context of an event and several candidate follow-up events, select the only correct follow-up event from these follow-up events. However, this event prediction method is not suitable for situations where there is an implicit causal relationship between some events. For example, the event pair for which there is no word related to "causality" in the text does not have a reasonable application to the situation where there is an implicit relationship. The causal relationship exists between events, but the possibility of the resulting event and the causal event appearing in the same context is judged by the co-occurrence probability between the events, so that the predicted event may appear inconsistent with the causal event. As a result, the predicted event description text is unreasonable, and the reference is not high.

SUMMARY OF THE INVENTION

The main purpose of this application is to provide an event description text prediction method, apparatus, device and computer-readable storage medium, aiming to improve the rationality and accuracy of the predicted event description text.

In a first aspect, the present application provides a description text prediction method for an event, and the event description text prediction method includes the following steps:

Obtain the event description text of the target event from the source terminal;

Semantically parse the event description text to obtain key verbs and key nouns;

Based on the event relationship tree, determine the degree of matching between the key verb and the node in the event relationship tree, wherein the node is used to indicate the first text and the second text corresponding to the first text, and the The first text is the cause event description text, and the second text is the result event description text;

According to the first node matching degree, determine the target node in the event relationship tree, and obtain the target first text and the target second text indicated by the target node;

Based on a preset noun relation database, determining the first degree of association between the nouns in the target first text and the nouns in the target second text;

Based on a preset semantic network, according to the first degree of association and the key noun, determine the target noun corresponding to the key noun, wherein the second degree of association between the target noun and the key noun is greater than or equal to the first degree of association;

Perform noun replacement processing on the target second text according to the target noun to obtain target description text, and send the target description text to the target terminal.

In a second aspect, the present application also provides a description text prediction device for an event, and the event description text prediction device includes:

The text acquisition module is used to acquire the event description text of the target event from the source terminal;

A semantic parsing module is used to perform semantic parsing on the event description text to obtain key verbs and key nouns;

The first node matching degree determination module is used to determine the first node matching degree between the key verb and the node in the event relationship tree based on the event relationship tree, wherein the node is used to indicate the first text and the The second text corresponding to the first text, the first text is the cause event description text, and the second text is the result event description text;

The target node determination module is used to determine the target node in the event relationship tree according to the first node matching degree, and obtain the target first text and the target second text indicated by the target node;

The first degree of association determination module is used to determine the first degree of association between nouns in the target first text and nouns in the target second text based on a preset noun relation database;

A target noun determination module is configured to, based on a preset semantic network, determine a target noun corresponding to the key noun according to the first degree of association and the key noun, wherein the target noun and the key noun are the same as the key noun. The second degree of association is greater than or equal to the first degree of association;

A target description text determination module, configured to perform a noun replacement process on the target second text according to the target noun to obtain a target description text, and send the target description text to the target terminal.

In a third aspect, the present application also provides a computer device comprising a processor, a memory, and a computer program stored on the memory and executable by the processor, wherein the computer program is executed by the When executed by the processor, the steps of the method for describing the text prediction as described above are realized.

In a fourth aspect, the present application further provides a computer-readable storage medium on which a computer program is stored, wherein when the computer program is executed by a processor, the above-mentioned method for predicting a description text of an event is implemented A step of.

The present application provides an event description text prediction method, device, device and computer-readable storage medium. The method includes acquiring an event description text of a target event from a source terminal; performing semantic analysis on the event description text to obtain key verbs and key nouns; Based on the event relationship tree, determine the first node matching degree between the key verb and the node in the event relationship tree, wherein the node is used to indicate the first text and the second text corresponding to the first text, and the first text is the cause event description text, The second text is the result event description text; according to the matching degree of the first node, the target node is determined in the event relationship tree, and the target first text and the target second text indicated by the target node are obtained; based on the preset noun relation database, determine The first degree of association between the nouns in the target first text and the nouns in the second target text; based on the preset semantic network, according to the first association degree and key nouns, determine the target nouns corresponding to the key nouns, where the target noun and the key noun are determined. The second association degree of the noun is greater than or equal to the first association degree; the target second text is subjected to noun replacement processing according to the target noun to obtain the target description text, and the target description text is sent to the target terminal. The present application determines the target node associated with the target event in the event relationship tree, and determines the association between the nouns in the text in the reason description text and the result description text contained in the target node, so that the event can be preserved when predicting the target event. Therefore, the rationality and reference of the predicted event description text can be improved.

Description of drawings

In order to explain the technical solutions of the embodiments of the present application more clearly, the following briefly introduces the accompanying drawings used in the description of the embodiments. For those of ordinary skill, other drawings can also be obtained from these drawings without any creative effort.

1 is a schematic flowchart of a method for predicting a description text of an event according to an embodiment of the present application;

2 is a schematic diagram of a scenario for implementing the method for predicting the description text of an event provided by the present application;

3 is a schematic diagram of another scenario for implementing the method for predicting the description text of an event provided by the present application;

4 is a schematic block diagram of an event description text prediction apparatus provided by an embodiment of the present application;

FIG. 5 is a schematic structural block diagram of a computer device according to an embodiment of the present application.

Detailed ways

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative work, all belong to the scope of protection of this application.

The flow charts shown in the figures are illustrative only and do not necessarily include all contents and operations/steps, nor do they have to be performed in the order described. For example, some operations/steps can be decomposed, combined or partially combined, so the actual execution order may change according to the actual situation.

Embodiments of the present application provide a description text prediction method, apparatus, computer device, and computer-readable storage medium for an event. The method for predicting the description text of the event can be applied to a terminal device, and the terminal device can be an electronic device such as a tablet computer, a notebook computer, and a desktop computer. It can also be applied to a server. The server can be a separate server, or it can provide cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and cloud servers for basic cloud computing services such as big data and artificial intelligence platforms.

Some embodiments of the present application will be described in detail below with reference to the accompanying drawings. The following embodiments and features of the embodiments may be combined with each other without conflict.

Please refer to FIG. 1, which is a schematic flowchart of a method for predicting a description text of an event according to an embodiment of the present application.

As shown in FIG. 1 , the method for predicting the description text of the event includes steps S101 to S107 .

Step S101 , acquiring the event description text of the target event from the source terminal.

Exemplarily, as shown in FIG. 2 , in the terminal, the description text of the target event can be generated in response to the user's input operation, so that the server or the terminal performing text prediction can obtain the description text of the target event from the source terminal. And after the server or the terminal that performs text prediction predicts and obtains the target description text, the target description text is sent to the target terminal.

Exemplarily, the target event may be a certain disaster event or a causal event that will cause another event to occur, and the description text may be used to describe this event. It can be understood that the description text includes multiple events. It can be understood that the description text of the target event includes but is not limited to Chinese or English.

Exemplarily, the description text of the target event can be obtained by voice conversion. Specifically, the description voice used by the user to describe the target event is obtained, and the text conversion model is trained to convert the description voice into text, so as to obtain the description voice of the target event. Description text. It can be understood that the speech can be converted into text in the source terminal, so that the server or the terminal performing text prediction can obtain the description text of the target event from the source terminal.

It can be understood that the words in the description text of the target event include various parts of speech, which may be, for example, nouns, verbs, and prepositions. Specifically, the description text of the target event may be, for example, that an earthquake occurs in country A, and the description text of the target event Among them, country A is a noun and an earthquake is a verb. And it can predict the disaster events or other events that will be affected by this earthquake by predicting the description text of the target event "A country has an earthquake". It can be understood that the foregoing description text of the target event is only an example, and does not limit the description text of the target event.

Step S102: Perform semantic analysis on the event description text to obtain key verbs and key nouns.

Exemplarily, the event description text may be input into the semantic parsing model, so that the semantic parsing model performs semantic parsing on the description text of the target event to obtain multiple keywords, and determines the part of speech of each keyword to obtain key verbs. and key nouns.

For example, in the semantic parsing model, word segmentation is performed on the description text of the target event by the word segmentation sub-model of the semantic parsing model to obtain at least two keywords; and the part of speech of each keyword is determined by the part of speech determination sub-model of the semantic parsing model, Get key verbs and key nouns.

Exemplarily, in the part-of-speech determination sub-model, the part-of-speech of each keyword can be determined by using a preset grammar template. The part-of-speech of each keyword can be determined in the matching between the grammar template and the keyword with word order.

Exemplarily, the preset grammar template may be a grammar template that focuses on verbs. Specifically, verbs can be identified through a VevbNet (verb network) vocabulary, and different verbs correspond to different preset grammar templates. It can be understood that in VevbNet ( After the verbs in the description text are identified by the verb network) vocabulary, the parts of speech corresponding to the remaining keywords in the description text are identified through the corresponding preset grammar templates. For example, in a verb grammar template, the part of speech of the word before the verb is a noun, and the part of speech of the word after the verb is an adverb of degree, then it can be determined that in the sentence matching the verb grammar template, before the key verb A word is a key noun, and a word after a key verb is an adverb of degree.

Matching the description text of the target event through the verb grammar template can improve the accuracy of identifying key verbs and key nouns.

Step S103, based on the event relationship tree, determine the first node matching degree between the key verb and the node in the event relationship tree, wherein the node is used to indicate the first text and the second corresponding to the first text. text, the first text is the cause event description text, and the second text is the result event description text.

Exemplarily, the event relationship tree may have a tree-like structure. It can be understood that the event relationship tree includes a root node and multiple child nodes of the root node, wherein the child nodes of multiple root nodes have the same hierarchical relationship, and each A node can be used to indicate the description text of a causal event pair. The description text of the causal event pair includes a first text for describing a causal event and a second text for describing a resultant event corresponding to the causal event, for example, the first text is: an earthquake occurs in country B, and the second text is: region C (the capital of country B) was devastated.

Exemplarily, the key verb may be matched with the verb in the first text indicated by the node in the event relationship tree, so as to obtain the first node matching degree between the key verb and the node of the event relationship tree. And through the first node matching degree, the first text matching the description text of the target event is determined in the event relation tree.

For example, the first text in the node with the highest matching degree of the first node is determined as the first text matching the description text of the target event.

In some embodiments, determining the first degree of matching between key verbs and nodes in the event relationship tree based on the event relationship tree includes: acquiring verbs in the first text corresponding to each node of the event relationship tree; based on a preset cosine similarity A gender algorithm is used to calculate the similarity between the key verb and the verb in the first text, to obtain the similarity between the key verb and each first text, and use the similarity as the matching degree of the first node between the keyword and the node.

Exemplarily, traverse the nodes of the event relationship tree, obtain the verbs in the first text indicated by each node, and perform similarity calculation between the obtained verbs of the first text and the key verbs to obtain the relationship between the key verbs and the first text. The similarity of verbs is used to determine the degree of matching between the key verb and each node of the event relation tree.

Exemplarily, based on a preset cosine similarity algorithm, calculating the similarity between the key verb and the verb in the first text may be, for example, performing vectorization processing on the key verb to obtain a key verb vector, and calculating the similarity between the key verb and the verb in the first text. The verbs of are vectorized to obtain the first text verb vector, and the similarity between the key verb vector and the first text verb vector is calculated according to the following formula:

Among them, A _i is used to indicate the key verb vector, B _i is used to indicate the first text verb vector, i is the vector element at the ith position in the vector, n is the number of vector elements in the vector, and cosθ is used to indicate the similarity Spend.

Through the above preset cosine similarity algorithm, the similarity between the key verb and the verb in the first text indicated by each node can be obtained, so as to determine the first matching degree between the key verb and each node of the event relationship tree.

Step S104, according to the matching degree of the first node, determine the target node in the event relationship tree, and obtain the target first text and the target second text in the target node.

Exemplarily, after calculating the first matching degree between the keyword and each node, the target node corresponding to the key verb is determined in the event relationship tree according to the first matching degree, so as to obtain the target first text and the target first text indicated by the target node. Target second text.

Understandably, the node with the highest first matching degree is used as the target node of the key verb.

Exemplarily, if the similarity between the key verb and the verb in the first text indicated by the multiple nodes is the same, then through the preset embedding network, determine the correlation strength between the cause event and the result event indicated by the node, specifically, The correlation strength can be used to indicate the probability of the occurrence of the consequential event after the occurrence of the causal event. If the probability of the occurrence of the consequential event after the occurrence of the causal event is 1, the weight of the node is determined to be 1; The probability is 50%, and the weight value of the node is determined to be 50%, so that the target node can still be determined when the verb similarity is the same. Understandably, the probability of occurrence can be determined according to historically occurring causal events and historically occurring resultant events.

The weight value of the matching degree of the first node is determined by the correlation strength of the cause event and the result event. The similarity between the key verb and the verb in the first text indicated by the node is multiplied by the corresponding weight value to obtain the first node matching degree to determine the target node.

In some embodiments, the event relationship tree includes at least two levels, and determining the target node in the event relationship tree according to the first node matching degree includes: determining the key verb and each node of the Nth level of the event relationship tree. A node matching degree, where N is a natural number greater than 0 and N is not greater than the total number of layers of the event relationship tree; the node whose first node matching degree is greater than the node matching threshold is determined as the node to be calculated; when N is less than the event relationship When the total number of layers of the tree, determine the degree of matching between the key verb and the second node of the child node of the node to be calculated; if the second degree of matching of the child node of the node to be calculated is smaller than the first As a matching degree, the node to be calculated is determined as the target node.

Exemplarily, the event relationship tree is a tree structure, including at least two layers, and except for the layer where the root node of the event relationship tree is located, the other layers include at least two nodes. It can be understood that the event relationship tree All nodes except the root node are child nodes of the root node.

Exemplarily, the root node is located at the first layer, the child nodes of the root node are the first child node and are located at the second layer, the child nodes of the first child node are the second child node and are located at the third layer, and so on to constitute an event. A relational tree, wherein the total number of layers of the event relational tree is M.

Exemplarily, the key verbs can be calculated from the Nth level of the event relationship tree, and the similarity can also be calculated from the root node of the event relationship tree. It is understandable that when the similarity calculation of the key verb starts from the Nth layer of the event relationship tree, the matching degree of the key verb and the first node of each node in the Nth layer of the event relationship tree is obtained, and the first node matching degree of each node is obtained. A node matching degree is compared with a preset node matching threshold, and a node whose first node matching degree is greater than the preset node matching threshold is regarded as the node to be calculated, and the matching degree is calculated for the child nodes of the node to be calculated.

It can be understood that when N is less than M, the calculation of the matching degree is performed on the child nodes of the node to be calculated, wherein the child nodes of the node to be calculated are located at layer N+1. When N is equal to M, the node to be calculated has no corresponding child node, and the node to be calculated is determined as the target node. It can be understood that if the node to be calculated has no child nodes, and the first nodes of at least two nodes to be calculated have the same degree of matching, the weight value can be calculated by the above-mentioned correlation strength between the cause event and the result event indicated by the node to be used in the calculation. Determine the target node among the nodes to be calculated.

Exemplarily, if the second degree of matching between the key verb and the child node of the node to be calculated is less than the first degree of matching, the target node is determined according to the node to be calculated, for example, the node to be calculated with the highest first degree of matching is determined, and is determined as the target node. .

If the second matching degree of the child nodes of the node to be calculated is greater than the first matching degree of the node to be calculated, add 1 to N.

Exemplarily, if the second matching degree of the child node of the node to be calculated is greater than the first matching degree, add 1 to N, so that the key verb continues to calculate the matching degree with the node in the event relation tree until N=M or If the second matching degree of the child nodes of the node to be calculated is smaller than the first matching degree, it is determined that the target node ends through the above steps, which will not be described here.

In some embodiments, the key verb is in the N+1 layer, and the matching degree is calculated by the key verb and the child node to be calculated, that is, in the N+1 layer of the event relationship tree, except for the child node to be calculated, the rest No node is calculated until the second matching degree is calculated to the M level of the event relationship tree or the child node to be calculated is smaller than the first matching degree, the target node is determined through the above steps, and the calculation is stopped.

In some other implementations, the key verbs and all the nodes in the N+1 layer are used to calculate the matching degree, and it is determined whether there is a third matching degree corresponding to the node that is greater than the second matching degree of the child nodes of the node to be calculated; if If it does not exist and N+1 is less than M, calculate the matching degree between the key verb and the child nodes of the node to be calculated; if it exists, determine the node whose third matching degree is greater than the second matching degree as the child node to be calculated, and calculate the child node to be calculated. The specific steps of the fourth matching degree corresponding to the parent node of the node and the sixth matching degree of the child node are as described above, and will not be described here.

By determining the target node by the matching degree, the accuracy of determining the target node can be improved.

Step S105: Determine the first degree of association between the nouns in the target first text and the nouns in the target second text based on a preset noun relation database.

Exemplarily, after the target node is determined, according to the first text and the second text indicated by the target node, and based on the noun relation database, the first noun in the first text indicated by the target node and the first noun in the second text are determined. The first degree of association between two nouns, it can be understood that the degree of association can be characterized by association information, including but not limited to the association information of geographic locations, person association information, and association information between items, for example, used to indicate parent-child association information The noun association degree is greater than the noun association degree used to indicate uncle and nephew association information. It can be understood that in practical applications, the association degree corresponding to the association information between nouns can be set according to requirements, which is not limited in this application.

In some embodiments, determining the first degree of association between the nouns in the target first text and the nouns in the target second text based on a preset noun relation database includes: obtaining the first degree of association between the nouns in the target first text a noun, and obtain a second noun in the target second text; according to the first noun, determine multiple noun triples to be selected in the noun relational database, wherein the noun triples to be selected are group contains the same noun as the first noun; determine whether the second noun exists in the candidate noun triplet; determine the first noun from the candidate noun triplet in which the second noun exists degree of association.

Exemplarily, the noun relation library includes multiple noun triples (v ₁ , v ₂ , l), where v ₁ is used to indicate a noun, v ₂ is used to indicate a noun that is related to v ₁ , and l is used to indicate a noun. It is used to indicate the association information of v ₁ and v ₂ , for example, v ₁ is the country of A, v ₂ is the region of B, and l is the capital of the country.

Exemplarily, the first noun is obtained from the first text indicated by the target node, and the second noun is obtained from the second text indicated by the same target node, and the noun relation database is searched for the relationship between the first noun and the second noun. Noun triples with the same noun, and the first degree of association between the first noun and the second noun is determined based on association information in the noun triple.

For example, a number of triples to be selected are determined in the noun relational database by the first noun. It is understandable that one of the nouns in the triplet to be selected is the same as the first noun, because one noun may have different names from multiple nouns. Therefore, the target triplet is determined from the plurality of candidate triples by the second noun, and the triplet with the same noun as the first noun and the second noun is determined as the target triplet. And obtain the first association information of the first noun and the second noun in the target triplet, and determine the first association degree of the first noun and the second noun based on the mapping relationship between the preset association information and the association degree.

Exemplarily, the degree of association between nouns can be determined by the association information of noun triples in the noun relational database, so as to improve the accuracy of target noun determination, thereby improving the rationality of the predicted target description text.

Step S106 , based on the preset semantic network, according to the first degree of association and the key noun, determine the target noun corresponding to the key noun, wherein the second degree of association between the target noun and the key noun greater than or equal to the first degree of association.

Exemplarily, the preset semantic network can be, for example, the open source Wordnet, and the target noun corresponding to the key noun can be determined by the key noun and the first degree of association, wherein the second degree of association between the key noun and the target noun is greater than The first degree of association, for example, the first association information is the capital of the country, it can be determined that the first association degree is level 5, and the association degree of important cities of the country is level 4, therefore, the second association degree needs to be greater than or equal to the first association degree. The degree, that is, the second association information corresponding to the second association degree is also used to indicate the capital of the country, and may not be the association information used to indicate the important city of the country.

For example, if the key noun is country C and the first related information is the capital of the country, the target noun should be the capital D region of country C. It is understandable that the key noun can also be other nouns, such as the noun indicating family relationship father, the first , and the second related information can also be other relationships, such as family relationships, and the key nouns, the first and second related information, and the target noun are not limited here.

Step S107: Perform noun replacement processing on the target second text according to the target noun to obtain a target description text, and send the target description text to the target terminal.

Exemplarily, after the target noun is determined, the noun in the second text indicated by the target node is replaced to obtain the target description text, wherein the target description text is used to describe the result event corresponding to the target event. And send the target description text to the target terminal.

For example, the target event is the earthquake in country C, and the first text indicated by the target node matching the target event is the earthquake in the country A, and the second text is the disaster in the region B (the capital of the country A); the target noun is determined as the country C through the semantic network. The capital D area of is obtained; obtain the second text indicated by the target node: area B is affected by disaster; replace the noun in the second text with the target noun, and get the target description text: area D is affected by disaster; thus complete the result event corresponding to the target event. predict. The obtained target description text is sent to the target terminal, so that the corresponding personnel can know the predicted target description text, so as to obtain the corresponding information, so as to set the corresponding preventive policy or execute the corresponding response policy.

In some embodiments, the target description text can also be stored in the blockchain, so that when the target terminal device needs to obtain the target description text, it can obtain the target description text by broadcasting to the blockchain. The blockchain referred to in this application is a new application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm. Blockchain, essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information to verify its Validity of information (anti-counterfeiting) and generation of the next block. The blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.

In some embodiments, before the determining the degree of matching between the key verb and the first node of the nodes in the event relationship tree based on the event relationship tree, the method further includes: acquiring a description text data set of sample event pairs, wherein, The sample event pair includes sample first text and sample second text corresponding to the sample first text, the sample first text is the sample cause event description text, and the sample second text is the sample result event description text; according to The sample first text and the sample second text included in the sample event pair are clustered on a plurality of the sample event pairs to obtain the event relationship tree.

Exemplarily, an event relationship tree is established, so that the first text matching the target event description text can be determined in the event relationship tree through the key verbs of the target event description text. Specifically, the description text data set of the sample event pair can be used, Build an event relationship tree. It can be understood that the description text dataset of sample event pairs includes description texts of multiple causal event pairs, wherein the causal event pairs include sample first text and sample second text, and the sample first text is the sample causal event description text. , the second sample text is the sample result event description text of the result event corresponding to the sample cause event.

Exemplarily, a plurality of event pairs are clustered to obtain nodes in an event relationship tree, thereby obtaining an event relationship tree. It can be understood that each cluster can be regarded as a node, and the event pair before the clustering can also be regarded as a node, so that multiple nodes can be obtained, and an event relationship tree can be obtained through the construction of multiple nodes.

Understandably, the established event key tree can also be stored in the blockchain.

In some embodiments, according to the sample first text and the sample second text included in the sample event pair, the plurality of sample event pairs are clustered to obtain the event relationship tree, including: Perform vectorization processing on the sample first text to obtain a first sample text vector, and perform vectorization processing on the sample second text to obtain a second sample text vector; based on the preset vector distance calculation formula, according to each The first sample text vector and the second sample text vector corresponding to the sample event pairs, determine the vector distance between the sample event pairs; The sample event pairs are classified into multiple first clusters of event pairs, wherein each of the first clusters of event pairs includes at least two sample event pairs; the vector distance between each of the first clusters of the event pairs is calculated, and according to the vector distance between the first clusters of the event pairs, classifying the first clusters of the pairs into a plurality of second clusters of the event pairs, wherein each of the second clusters of the event pairs at least Two first clusters of the event pairs are included; the event relationship tree is determined according to each of the first clusters of the event pairs and each of the second clusters of the event pairs.

Exemplarily, performing vectorization processing on the sample first text and the sample second text in each sample event pair to obtain the corresponding sample first text vector and sample second text vector, and based on the preset vector distance calculation formula, Calculate the vector distance between the sample first text vector and the sample second text vector of different sample event pairs. It is understandable that the distance between the sample first text vector corresponding to different sample event pairs and the difference between the sample second text vector The sum of the distances can represent the vector distance between different sample event pairs, as shown in the following formula:

SIM(<c _i ,e _i >,<c _j ,e _j >)=SIM(c _i ,c _j )+SIM(e _i ,e _j )

Among them, SIM is used to indicate the vector distance, c _i represents the sample first text vector of the ith sample event pair, e _i represents the sample second text vector of the ith sample event pair; c _j represents the j th sample event pair The sample first text vector of , e _j represents the sample second text vector of the jth sample event pair.

It can be understood that after calculating the vector distances between different sample event pairs, multiple sample event pairs can be classified and processed through the vector distances between different sample event pairs, as shown in Figure 3, to obtain a plurality of first clustering; and calculating the vector distance between the plurality of first clusters, and classifying the plurality of first clusters to obtain a plurality of second clusters, and by analogy, to establish an event relationship tree.

In some embodiments, classifying each of the sample event pairs into a first cluster of a plurality of event pairs according to a vector distance between each of the sample event pairs includes: determining a first sample event pair, and Taking the sample event pairs other than the first sample event pair in the descriptive text data set of the current event pair as the target sample event pair in turn; calculating the first sample event pair and the target sample event pair Determine the target sample event pair with the smallest vector distance from the first sample event pair as the second sample event pair, and classify the first sample event pair and the second sample event pair into The same event pair is in the first cluster.

Exemplarily, two sample event pairs with the smallest vector distance are determined, and clustering is performed to obtain the first cluster of event pairs; specifically, one sample event pair is set as the target sample event pair, and the remaining sample event pairs and the target are calculated. The vector distance between the sample event pairs, determine the sample event pair with the smallest vector distance from the target sample event pair among the remaining sample event pairs, and cluster the sample event pair with the target sample event pair to obtain the first event pair. Clustering; in the remaining sample event pairs after clustering, the target sample event pair is determined again, and the above calculation is repeated to obtain a first cluster of multiple event pairs.

Exemplarily, the sample first text vector and the sample second text vector corresponding to the sample event included in the first cluster of the event pair are added to obtain the sample third text vector and the sample corresponding to the first cluster of the event pair. The fourth text vector, and the vector distance between the first clusters of each event pair is calculated through the sample third text vector and the sample fourth text vector corresponding to the first cluster of the event pair, so as to determine the second cluster of the event pair, with This calculation method is analogous until the Nth cluster of event pairs is obtained, which can include all sample event pairs in the sample event pair dataset. It can be understood that N is also used to indicate the number of layers of the event relationship tree.

Exemplarily, the event pair Nth cluster can be used to indicate the node of the event relationship tree. It can be understood that the first text and the second text indicated by the node should be able to include the first text and the second text of the child nodes of the node. The information indicated by the text, therefore, when the event pair Nth cluster is obtained, it is necessary to generalize the first text and the second text contained in the Nth cluster to determine the Nth cluster, that is, the node corresponding to the first text and the second text.

Specifically, based on the word segmentation model, the first text of the sample can be segmented to obtain multiple keywords; based on the semantic generalization model, the semantic generalization of each keyword can be performed to obtain the attribute concept word corresponding to each keyword. , where the attribute concept word is used to describe the attribute or category of the keyword; based on the entity word database, entity matching is performed on the attribute concept word to obtain the entity word corresponding to the attribute concept word, and the entity word is the attribute or category of the attribute concept word. associated words; splicing each entity word to obtain the first text indicated by the parent node of the node where the first text of the sample is located; similarly, the second text of the sample can be processed in the same way to obtain the The second text indicated by the parent node is obtained, thereby obtaining the first text and the second text indicated by the node corresponding to the Nth cluster of the event, and establishing an event relation tree through multiple nodes.

The method for predicting the description text of the event provided by the above-mentioned embodiment, determines the first text matching the description text of the target event in the event relation tree, and uses the second text corresponding to the first text, the first text and the nouns in the second text. Predict the target description text of the result event corresponding to the target event, so that the existing relationship between the events can be preserved, thereby improving the rationality and reference of the predicted target description text.

Please refer to FIG. 4 . FIG. 4 is a schematic diagram of an event description text prediction device provided by an embodiment of the present application. The event description text prediction device may be configured in a server or a terminal for executing the aforementioned event description text. method of prediction.

As shown in FIG. 4, the description text prediction device 100 of the event includes: a text acquisition module 110, a semantic analysis module 120, a first node matching degree determination module 130, a target node determination module 140, a first correlation degree determination module 150, The target noun determination module 160 and the target description text determination module 170 .

The text obtaining module 110 is configured to obtain the event description text of the target event from the source terminal.

The semantic parsing module 120 is configured to perform semantic parsing on the event description text to obtain key verbs and key nouns.

The first node matching degree determination module 130 is configured to determine, based on the event relationship tree, the first node matching degree between the key verb and the node in the event relationship tree, wherein the node is used to indicate the first text and the second text corresponding to the first text, the first text is the cause event description text, and the second text is the result event description text;

The target node determination module 140 is configured to determine a target node in the event relationship tree according to the matching degree of the first node, and obtain the target first text and the target second text indicated by the target node.

The first association degree determination module 150 is configured to determine the first association degree between the nouns in the target first text and the nouns in the target second text based on a preset noun relation database.

A target noun determination module 160, configured to determine a target noun corresponding to the key noun according to the first degree of association and the key noun based on a preset semantic network, wherein the target noun and the key noun The second degree of association is greater than or equal to the first degree of association.

The target description text determination module 170 is configured to perform noun replacement processing on the target second text according to the target noun to obtain target description text, and send the target description text to the target terminal.

Exemplarily, the first node matching degree determination module 130 further includes a verb acquisition sub-module and a similarity degree calculation sub-module.

The verb acquisition submodule is used to acquire verbs in the first text corresponding to each node of the event relationship tree.

A similarity calculation sub-module, used for calculating the similarity between the key verb and the verb in the first text based on a preset cosine similarity algorithm, to obtain the similarity between the key verb and each of the first texts degree, and use the similarity degree as the first node matching degree of the key verb and the node.

Exemplarily, the target node determination module 140 is further configured to:

Determine the first node matching degree of the key verb and each node of the Nth layer of the event relationship tree, wherein, N is a natural number greater than 0 and N is not greater than the total number of layers of the event relationship tree;

Determining a node whose first node matching degree is greater than the node matching threshold as a node to be calculated;

When N is less than the total number of layers of the event relationship tree, determine the second node matching degree of the key verb and the child node of the node to be calculated;

If the second matching degree of the child node of the node to be calculated is less than the first matching degree of the node to be calculated, determine the target node according to the node to be calculated;

If the second matching degree of the child nodes of the node to be calculated is greater than the first matching degree of the node to be calculated, add 1 to N.

Exemplarily, the first association degree determination module 150 includes a noun acquisition submodule, a noun triplet determination submodule, and a second noun judgment submodule.

A noun acquisition submodule for acquiring the first noun in the first text of the target, and acquiring the second noun in the second text of the target;

A noun triplet determination submodule is used to determine a plurality of noun triples to be selected in the noun relation database according to the first noun, wherein the noun triples to be selected include the same as the first noun. nouns with the same noun;

The second noun judging submodule is used to determine whether the second noun exists in the noun triplet to be selected;

The first degree of association determination module 150 is further configured to determine the first degree of association from the candidate noun triples in which the second noun exists.

Exemplarily, the event description text prediction apparatus 100 further includes a sample description text acquisition module and an event relationship tree establishment module.

The sample description text acquisition module is used to obtain the description text data set of the sample event pair, wherein the sample event pair includes the sample first text and the sample second text corresponding to the sample first text, and the sample first text is Sample cause event description text, the second sample text is sample result event description text;

The event relationship tree building module is used for clustering a plurality of the sample event pairs according to the sample first text and the sample second text included in the sample event pair to obtain the event relationship tree.

Exemplarily, the event relationship tree building module includes a vector processing submodule, a vector distance calculation submodule, an event pair classification submodule, and a cluster classification submodule.

The vector processing submodule is configured to perform vectorization processing on the first text of the sample to obtain the first text vector of the sample, and perform vectorization processing on the second text of the sample to obtain the second text vector of the sample.

The vector distance calculation submodule is used to determine the vector distance between each described sample event pair based on the preset vector distance calculation formula, according to the sample first text vector and the sample second text vector corresponding to each described sample event pair;

The event pair classification submodule is configured to classify each of the sample event pairs into a first cluster of a plurality of event pairs according to the vector distance between the sample event pairs, wherein each of the event pairs is the first cluster The class includes at least two sample event pairs;

A cluster classification sub-module is used to calculate the vector distance between each of the event pairs of the first cluster, and according to the vector distance between each of the event pairs of the first cluster, classify each of the pairs of the first cluster Classifying into a plurality of second clusters of event pairs, wherein each of the second clusters of event pairs includes at least two of the first clusters of event pairs;

The event relationship tree establishing module is further configured to determine the event relationship tree according to the first clustering of each of the event pairs and the second clustering of each of the event pairs.

Exemplarily, the event pair classification submodule further includes a target sample event pair determination submodule, an event pair vector distance calculation submodule, and an identification comparison submodule.

An identification determination submodule is used to determine a first sample event pair, and the sample event pairs other than the first sample event pair in the description text data set of the current event pair are sequentially used as the target sample event pair.

A language identification determination sub-module for calculating the vector distance between the first sample event pair and the target sample event pair.

The event pair classification submodule is further configured to determine the target sample event pair with the smallest vector distance from the first sample event pair as the second sample event pair, and combine the first sample event pair with the second sample event pair The sample event pairs are classified into the same event pair first cluster.

It should be noted that those skilled in the art can clearly understand that, for the convenience and brevity of description, for the specific working process of the above-described device and each module and unit, reference may be made to the corresponding process in the foregoing method embodiments. No longer.

The methods of the present application can be used in numerous general purpose or special purpose computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, including A distributed computing environment for any of the above systems or devices, and the like. The application may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The application may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including storage devices.

Please refer to FIG. 5. FIG. 5 is a schematic block diagram of the structure of a computer device according to an embodiment of the present application. The computer equipment can be a server or a terminal.

As shown in Figure 5, the computer device includes a processor, memory, and a network interface connected by a system bus, wherein the memory may include a storage medium and an internal memory.

The storage medium may store an operating system and a computer program. The computer program includes program instructions that, when executed, can cause the processor to execute any description text prediction method for events.

The processor is used to provide computing and control capabilities to support the operation of the entire computer equipment.

The internal memory provides an environment for running a computer program in the storage medium, and when the computer program is executed by the processor, the processor can cause the processor to execute any description text prediction method for events.

The network interface is used for network communication, such as sending assigned tasks. Those skilled in the art can understand that the structure shown in the figure is only a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer equipment to which the solution of the present application is applied. The specific computer equipment may include There are more or fewer components than shown in the figures, or some components are combined, or have a different arrangement of components.

It should be understood that the processor may be a central processing unit (Central Processing Unit, CPU), and the processor may also be other general-purpose processors, digital signal processors (Digital Signal Processors, DSP), application specific integrated circuits (Application Specific Integrated Circuits) Circuit, ASIC), Field-Programmable Gate Array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like. Wherein, the general-purpose processor can be a microprocessor or the processor can also be any conventional processor or the like.

Wherein, in one embodiment, described processor is used for running the computer program stored in the memory, to realize the following steps:

Obtain the event description text of the target event from the source terminal;

Semantically parse the event description text to obtain key verbs and key nouns;

Based on the event relationship tree, determine the degree of matching between the key verb and the node in the event relationship tree, wherein the node is used to indicate the first text and the second text corresponding to the first text, and the The first text is the cause event description text, and the second text is the result event description text;

According to the first node matching degree, determine the target node in the event relationship tree, and obtain the target first text and the target second text indicated by the target node;

Based on a preset noun relation database, determining the first degree of association between the nouns in the target first text and the nouns in the target second text;

Based on a preset semantic network, according to the first degree of association and the key noun, determine the target noun corresponding to the key noun, wherein the second degree of association between the target noun and the key noun is greater than or equal to the first degree of association;

Perform noun replacement processing on the target second text according to the target noun to obtain target description text, and send the target description text to the target terminal.

In one embodiment, when the processor determines the first node matching degree of the key verb and the node in the event relationship tree based on the event relationship tree, it is used to realize:

acquiring the verbs in the first text corresponding to each node of the event relationship tree;

Based on a preset cosine similarity algorithm, calculate the similarity between the key verb and the verb in the first text to obtain the similarity between the key verb and each of the first texts, and calculate the similarity between the key verbs and the first texts. As the key verb and the first node matching degree of the node.

In one embodiment, when the processor determines the target node in the event relationship tree according to the first node matching degree, it is used to realize:

Determine the first node matching degree of the key verb and each node of the Nth layer of the event relationship tree, wherein, N is a natural number greater than 0 and N is not greater than the total number of layers of the event relationship tree;

Determining a node whose first node matching degree is greater than the node matching threshold as a node to be calculated;

When N is less than the total number of layers of the event relationship tree, determine the second node matching degree of the key verb and the child node of the node to be calculated;

If the second matching degree of the child node of the node to be calculated is less than the first matching degree of the node to be calculated, determine the target node according to the node to be calculated;

If the second matching degree of the child nodes of the node to be calculated is greater than the first matching degree of the node to be calculated, add 1 to N.

In one embodiment, when the processor determines the first degree of association between the nouns in the target first text and the nouns in the target second text based on a preset noun relational database, the processor is configured to:

Obtain the first noun in the target first text, and obtain the second noun in the target second text;

According to the first noun, a plurality of noun triples to be selected are determined in the noun relation database, wherein the noun triples to be selected include the same noun as the first noun;

determining whether the second noun exists in the candidate noun triplet;

The first degree of association is determined from candidate noun triples in which the second noun exists.

In one embodiment, the processor is used to realize before determining the matching degree of the key verb and the first node of the node in the event relationship tree based on the event relationship tree:

Obtain a description text data set of a sample event pair, wherein the sample event pair includes a sample first text and a sample second text corresponding to the sample first text, the sample first text is a sample cause event description text, and the The second text of the sample is the description text of the sample result event;

According to the sample first text and the sample second text contained in the sample event pair, a plurality of the sample event pairs are clustered to obtain the event relationship tree.

In one embodiment, the processor implements clustering a plurality of the sample event pairs according to the sample first text and the sample second text included in the sample event pair to obtain the event relationship tree, used to implement:

The first text of the sample is vectorized to obtain the first text vector of the sample, and the second text of the sample is vectorized to obtain the second text vector of the sample;

Based on the preset vector distance calculation formula, according to the sample first text vector and the sample second text vector corresponding to each of the sample event pairs, determine the vector distance between each of the sample event pairs;

According to the vector distance between the sample event pairs, each of the sample event pairs is classified into a first cluster of multiple event pairs, wherein each of the first cluster of event pairs includes at least two sample event pairs ;

Calculate the vector distance between the first clusters of each of the event pairs, and classify each of the first pairs of clusters into a plurality of event pairs of the second cluster according to the vector distance between the first clusters of the event pairs. In the class, wherein, each of the event pair second cluster includes at least two of the event pair first cluster;

The event relationship tree is determined based on each of the event pairs of the first cluster and each of the event pairs of the second cluster.

In one embodiment, when the processor implements classifying each of the sample event pairs into a first cluster of multiple event pairs according to the vector distance between each of the sample event pairs, it is used to implement:

Determine the first sample event pair, and use the sample event pairs other than the first sample event pair in the description text data set of this event pair as the target sample event pair in turn;

Calculate the vector distance between the first sample event pair and the target sample event pair;

Determine the target sample event pair with the smallest vector distance from the first sample event pair as the second sample event pair, and classify the first sample event pair and the second sample event pair into the same event pair in the first cluster.

It should be noted that those skilled in the art can clearly understand that, for the convenience and brevity of the description, for the specific working process of the description text prediction of the above-mentioned events, you can refer to the corresponding descriptions in the description text prediction control method embodiments of the aforementioned events. The process is not repeated here.

Embodiments of the present application further provide a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, the computer program includes program instructions, and the method implemented when the program instructions are executed may refer to this document Various embodiments of description text prediction methods for application events.

Wherein, the computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiments, such as a hard disk or a memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) equipped on the computer device card, flash card (Flash Card) and so on.

It should be understood that the terminology used in the specification of the present application is for the purpose of describing particular embodiments only and is not intended to limit the present application. As used in this specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural unless the context clearly dictates otherwise.

It should also be understood that, as used in this specification and the appended claims, the term "and/or" refers to and including any and all possible combinations of one or more of the associated listed items. It should be noted that, herein, the terms "comprising", "comprising" or any other variation thereof are intended to encompass non-exclusive inclusion, such that a process, method, article or system comprising a series of elements includes not only those elements, It also includes other elements not expressly listed or inherent to such a process, method, article or system. Without further limitation, an element qualified by the statement "comprising a..." does not preclude the presence of additional identical elements in the process, method, article or system that includes the element.

The above-mentioned serial numbers of the embodiments of the present application are only for description, and do not represent the advantages or disadvantages of the embodiments. The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of various equivalents within the technical scope disclosed in the present application. Modifications or substitutions shall be covered by the protection scope of this application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims (10)

1. a description text prediction method of an event, is characterized in that, comprises: Obtain the event description text of the target event from the source terminal; Semantically parse the event description text to obtain key verbs and key nouns; Based on the event relationship tree, determine the degree of matching between the key verb and the node in the event relationship tree, wherein the node is used to indicate the first text and the second text corresponding to the first text, and the The first text is the cause event description text, and the second text is the result event description text; According to the matching degree of the first node, a target node is determined in the event relationship tree, and the target first text and the target second text indicated by the target node are obtained; determining the first degree of association between the nouns in the target first text and the nouns in the target second text based on a preset noun relation database; Based on a preset semantic network, according to the first degree of association and the key noun, determine the target noun corresponding to the key noun, wherein the second degree of association between the target noun and the key noun is greater than or equal to the first degree of association; Perform noun replacement processing on the target second text according to the target noun to obtain target description text, and send the target description text to the target terminal. 2. The method for predicting the description text of an event as claimed in claim 1, wherein the determining the degree of first node matching between the key verb and a node in the event relationship tree based on the event relationship tree comprises: acquiring the verbs in the first text corresponding to each node of the event relationship tree; Based on a preset cosine similarity algorithm, calculate the similarity between the key verb and the verb in the first text to obtain the similarity between the key verb and each of the first texts, and calculate the similarity between the key verbs and the first texts. As the key verb and the first node matching degree of the node. 3 . The method for predicting the description text of an event according to claim 1 , wherein the event relationship tree comprises at least two levels, and the event relationship tree is determined according to the first node matching degree in the event relationship tree. 4 . target node, including: Determine the first node matching degree between the key verb and each node of the Nth layer of the event relationship tree, wherein N is a natural number greater than 0 and N is not greater than the total number of layers of the event relationship tree; Determining a node whose first node matching degree is greater than the node matching threshold as a node to be calculated; When N is less than the total number of layers of the event relationship tree, determine the degree of matching between the key verb and the second node of the child node of the node to be calculated; If the second matching degree of the child nodes of the node to be calculated is smaller than the first matching degree of the node to be calculated, determine the target node according to the node to be calculated; If the second matching degree of the child nodes of the node to be calculated is greater than the first matching degree of the node to be calculated, add 1 to N. 4. The method for predicting the description text of an event according to claim 1, wherein, based on a preset noun relation database, determining the number of nouns in the target first text and the nouns in the target second text. a degree of association, including: Obtain the first noun in the target first text, and obtain the second noun in the target second text; According to the first noun, a plurality of noun triples to be selected are determined in the noun relation database, wherein the noun triples to be selected include the same noun as the first noun; determining whether the second noun exists in the candidate noun triplet; The first degree of association is determined from candidate noun triples in which the second noun exists. 5. The method for predicting the description text of an event according to any one of claims 1-3, wherein, in the event-based relationship tree, determining the key verb and the first node of the node in the event relationship tree Before match, also includes: Obtain a description text data set of a sample event pair, wherein the sample event pair includes a sample first text and a sample second text corresponding to the sample first text, the sample first text is a sample cause event description text, and the The second text of the sample is the description text of the sample result event; According to the sample first text and the sample second text contained in the sample event pair, a plurality of the sample event pairs are clustered to obtain the event relationship tree. 6 . The method for predicting the description text of an event according to claim 5 , wherein, according to the sample first text and the sample second text included in the sample event, for a plurality of the sample events. 7 . Perform clustering on the pair to obtain the event relationship tree, including: Perform vectorization processing on the sample first text to obtain a sample first text vector, and perform vectorization processing on the sample second text to obtain a sample second text vector; Based on a preset vector distance calculation formula, according to the sample first text vector and the sample second text vector corresponding to each sample event pair, determine the vector distance between each of the sample event pairs; According to the vector distance between the sample event pairs, each of the sample event pairs is classified into a first cluster of multiple event pairs, wherein each of the first cluster of event pairs includes at least two sample event pairs ; Calculate the vector distance between the first clusters of each of the event pairs, and classify each of the first pairs of clusters into a plurality of event pairs of the second cluster according to the vector distance between the first clusters of the event pairs. class, wherein each of the event pair second clusters includes at least two of the event pair first clusters; The event relationship tree is determined according to each of the first clusters of the event pairs and each of the event pairs to the second cluster. 7 . The method for predicting the description text of events according to claim 6 , wherein, according to the vector distance between the sample event pairs, the sample event pairs are classified into a plurality of event pairs first clustering. 8 . class, including: determining a first sample event pair, and sequentially taking the sample event pairs other than the first sample event pair in the description text data set of the current event pair as the target sample event pair; Calculate the vector distance between the first sample event pair and the target sample event pair; Determine the target sample event pair with the smallest vector distance from the first sample event pair as the second sample event pair, and classify the first sample event pair and the second sample event pair into the same event pair in the first cluster. 8. An event description text prediction device, wherein the event description text prediction device comprises: The text acquisition module is used to acquire the event description text of the target event from the source terminal; a semantic parsing module, used for semantic parsing of the event description text to obtain key verbs and key nouns; The first node matching degree determination module is used to determine the first node matching degree between the key verb and the node in the event relationship tree based on the event relationship tree, wherein the node is used to indicate the first text and the The second text corresponding to the first text, the first text is the cause event description text, and the second text is the result event description text; a target node determination module, configured to determine a target node in the event relationship tree according to the first node matching degree, and obtain the target first text and the target second text indicated by the target node; a first degree of association determination module, configured to determine a first degree of association between nouns in the target first text and nouns in the target second text based on a preset noun relation database; A target noun determination module is configured to, based on a preset semantic network, determine a target noun corresponding to the key noun according to the first degree of association and the key noun, wherein the target noun and the key noun are the same as the key noun. The second degree of association is greater than or equal to the first degree of association; A target description text determination module, configured to perform noun replacement processing on the target second text according to the target noun, obtain target description text, and send the target description text to the target terminal. 9. A computer device comprising a processor, a memory, and a computer program stored on the memory and executable by the processor, wherein the computer program is executed by the processor When executed, the steps of the method for descriptive text prediction of events as claimed in any one of claims 1 to 7 are implemented. 10. A computer-readable storage medium, characterized in that, a computer program is stored on the computer-readable storage medium, and when the computer program is executed by the processor, the computer program as claimed in any one of claims 1 to 7 is implemented. Describe the steps of the text prediction method for the described event.

Images