TWM523902U - Search engine device for collecting keyword - Google Patents

Description

Search engine device that can collect keywords

This creation relates to a search engine device, and more particularly to a search engine device that collects keyword keywords and provides a list of recommendations.

With the popularity of the Internet, search engines have become an indispensable technology in people's daily work and life, and the "Google Suggest" technology has attracted more attention. "Google Suggest" is that after the user enters the query keyword, the search engine will automatically generate a frame to display other commonly used related keywords before the search, as a list of recommendations for the user to further select. This technology not only reminds users of other highly relevant keywords, but also helps users quickly find the information they are looking for and improve their search efficiency.

Due to the convenience of online search, large-scale real estate agents have also begun to invest in the research and development of real estate search engines, so that users can quickly find the information they need from thousands of real estate materials. The technology of "Google Suggest" It has also become an important part of the real estate search engine. Due to the large number of real estate objects, the content information is also complicated, and the popularity of specific information will change with time. Therefore, it is easy to make some meanings to most users in the recommendation list based on the keyword relevance. The keywords are placed above the box, so the recommendation list can't fully recommend the current hottest keywords to meet the user's needs. For example, when When entering "Nangang", "Nangang National House", "Nangang Park", "Nangang Software Park", "Nangang Institute of Research", "Nangang Station", "Nangang Road Section", etc. appear in the list of recommendations from top to bottom. However, the “Nangang Road Section” that most users are most frequently searching for and most concerned about is listed below the recommendation list. Since the topmost keyword is usually the most noticeable and clicked by the user, if the popular keyword is not placed, In the prominent position in the recommendation list, it is easy to be ignored by the user or it takes more time to find, which makes the convenience of the recommendation list lower.

Therefore, how to collect popular keywords that users often use to produce more accurate and convenient recommendation lists has become an urgent issue in the field.

In view of the deficiencies in the prior art, the author of this creation has carefully tested and researched and has a perseverance spirit to finally conceive the creation of a search engine device that can collect keywords and overcome the deficiencies of the prior art. A brief description of this creation.

The purpose of this creation is to record the attributes selected by the user or the corresponding number of cumulative clicks, the cumulative number of searches and the number of associated real estate objects, based on the real estate object information, by the size of the attribute data and its changes. Identify the popularity and representativeness of the keyword, and add different weights to the keywords in the recommendation list to determine and instantly update the order of the keywords in the recommendation list to produce a more accurate and convenient recommendation list. Improve search efficiency.

The present invention provides a search engine device that can collect keywords, including: a database configured to store a plurality of first keywords, and each of the first keywords having an attribute data, the attribute data including at least one accumulated The number of times selected and the number of times the accumulated search is performed; an analysis module configured to analyze a string in a data source and analyze the string When there is real estate relevance, the string is defined as the first keyword and stored in the database; a related module is configured to establish the first number by using an inter-lexical correlation analysis technique Correlation between keywords, such that each first keyword is related to a plurality of second keywords, and is recorded in a related information, the plural second keyword is derived from the first keyword, and the first keyword is used a property data of a keyword; a sorting module configured to determine each first keyword according to the cumulative number of times of the plurality of the second keywords or/and the cumulative number of times of the search Corresponding one of each of the plurality of second keywords is initially arranged; a communication module receives a search string input from a user; and a search module configured to receive the search string and find Determining the first keyword that matches the search string and defining the second keyword by the relevance module to display the plural second keyword according to the initial ranking order determined by the sorting module for A list of recommendations.

1‧‧‧Search engine unit

10‧‧‧Database

11‧‧ Real estate object database

12‧‧‧Real estate related vocabulary database

15‧‧‧ field

16‧‧‧Search box

17‧‧‧ string

18‧‧‧Recommended list

20‧‧‧Analysis module

30‧‧‧Affinity module

35‧‧‧Sorting module

40‧‧‧Communication module

50‧‧‧Search Module

X ₀ ~X _n , Y ₀ ~Y ₂ ‧‧‧ keywords

D‧‧‧ semantic distance

The first figure is a block diagram of a search engine device of one embodiment of the present creation.

The second figure is a schematic diagram of the search state of one embodiment of the present creation.

The third figure is a schematic diagram of the selection list selection state of one embodiment of the present creation.

The fourth figure is a schematic diagram of the keyword tentative arrangement order of one embodiment of the present creation.

The fifth figure is a schematic diagram of a recommended list of one embodiment of the present creation.

The technical content, features, and effects of the present invention will be apparent from the following detailed description of the preferred embodiments.

Please refer to the first figure, which is a search engine device of one embodiment of the present creation. 1 schematic diagram. The search engine device 1 includes a database 10, an analysis module 20, an association module 30, a sorting module 35, a communication module 40, and a search module 50.

The database 10 further includes: a real estate object database 11 and a real estate related vocabulary database 12. The real estate object database 11 records a plurality of real estate object materials. In more detail, the records include object descriptions, addresses, community names, and/or object abbreviations. The real estate related vocabulary database 12 records a plurality of real estate related vocabulary, wherein the plurality of real estate related vocabularies are mainly based on the rule of thumb, and through statistical analysis, which nouns are related to real estate related vocabulary, for example: county name, city name, township Name, town name, district name, liname, street name, well-known landmark name, public facility name, known community name and known object short name are used as the analysis basis of analysis module 20. More specifically, the specific nouns in the real estate object database 11 appear particularly frequently, and the system will indicate real estate related vocabulary, otherwise it will not be input to the real estate object database 11 from the beginning.

The analysis module 20 of the present invention is for analyzing strings from data sources. The data source may be real estate object data recorded by the real estate object database 11 or the communication module 40 receives the string from the user. The analysis module 20 compares the strings from the real estate object database 11 or the communication module 40 according to the real estate related vocabulary recorded in the real estate related vocabulary database 12 to analyze whether the string has real estate relevance. When the string has real estate relevance, the analysis module 20 defines the string as a first keyword and stores the first keyword in the database 10, and the first keyword includes an address, a community name, or an object. Abbreviation.

The database 10 stores a plurality of first keywords, each of which has an attribute data. The attribute data includes at least one cumulative number of times selected and one accumulated number of times of searching. Please refer to the second figure, which is a schematic diagram of the search state of one embodiment of the present creation. In the second figure, when the "Daan" string 17 is entered in field 15, and then the search box 16 is selected, The accumulation of "Daan" of a keyword is increased by the number of searches and recorded in the database 10. Please refer to the third figure. The third figure is a schematic diagram of the recommended list point selection state in one embodiment of the present creation. In the third figure, when "Daan" is entered in field 15, a list of recommendations 18 is generated listing a plurality of recommended strings. After the user clicks on "Da'an National House", the cumulative number of clicks of "Da'an National House" as the first keyword will be increased once and recorded in the database 10.

The association module 30 is configured to establish an association (speech distance) between the plurality of first keywords by using an inter-lexical correlation analysis technique. The so-called inter-linguistic correlation analysis technique can be calculated by a semantic analysis algorithm such as a conventional KuLback-Leibler Divergence or a Normalize Google Distance. Here is an example of normalizing Google distance. The semantic distance formula is as follows.

In the above semantic distance formula, the number of co-occurrences X is searched for the total number of N real estate documents, searching for two keywords A and B, and obtaining the number of results of A and B co-occurring in the same real estate object data. The number of occurrences Y and Z are in the real estate object data of N, respectively searching for the two keywords A and B, and obtaining the number of second search results that A and B appear in the real estate object data. That is to say, if the semantic distance D of the keywords "Nangang" (A) and "South House" (B) is to be calculated, it is recorded in the real estate object database 11 via the association module 30, and the total number is 8,058,044,651 ( N) The number of "Southern Port" in the real estate related text is 46,700,000 (Y) and the number of "Southern House" is 12,200,000 (Z), and the number of occurrences of "Nangang" and "Southern House" is 2,630,000 (X). The above formula calculates two keywords The semantic distance D is 0.44305. The higher the value of the semantic distance D is, the lower the relevance of the two keywords is. The closer the D value is to zero, the higher the relevance of the two keywords, that is, the two keywords almost all appear at the same time. In addition, if the number of search results (Y or Z) is close to the total number N of searches, it means that the keyword is not representative, so the association module 30 removes the keyword whose number of search results is close to the total number of searches, without calculating the semantics. distance.

After calculating the association between the plurality of first keywords, the association module 30 defines the keyword related to the first keyword as the second keyword, and records the information in a related information, where the The second keyword is from the first keyword, and the attribute data possessed by the first keyword is used, so that the second keyword is compared and weighted later.

The association module 30 repeats the above manner, and can establish the association data between each of the first keyword and the second keyword according to the semantic distance. Then, the sorting module 35 sorts each of the first keyword and the second keyword (usually a plurality of characters) into a tentative arrangement order according to the semantic distance. Please refer to the fourth figure. The fourth figure is a schematic diagram of the keyword tentative arrangement order of one embodiment of the present creation. It can be seen from the fourth figure that the two tentative arrangement orders X ₀ ~X _n and Y ₀ ~Y ₂ which are respectively the vertices of the first keywords X ₀ and Y ₀ are not connected, which represents the processing by the association module 30. The keyword groups will also be completely unrelated to each other, so they belong to different provisional order. In the provisional arrangement order, any two keywords have their semantic distance, the semantic distance of the first keyword X ₀ and the second keyword X ₁ is D(X ₀ , X ₁ ), and the first keyword X ₀ and The semantic distance of the second keyword X ₂ is D(X ₀ , X ₂ ), the semantic distance of the first keyword X ₀ and the second keyword X _n is D(X ₀ , X _n ), and so on. In order to simplify the illustration, all semantic distances are not listed here. In addition, although in the fourth figure, the tentative order is a continuous column, this is only a display state, and is not limited to any arrangement, as long as the semantic distance between the keywords can be expressed. The sorting module 35 may include the second keyword group (ie, X ₁ ~X _n ) of the keyword association sequence X ₀ ~X _n with the first keyword "closer semantic distance" (highly correlated) The same provisional order. It should be specially pointed out that the so-called "closer semantic distance" refers to a situation in which the semantic distance is less than a preset distance threshold, so in a provisional arrangement order, the second keyword having a semantic distance less than the preset distance threshold usually has Multiple. The preset distance threshold can be appropriately set by those skilled in the art.

After the provisional ordering sequence is generated, the sorting module 35 adds the first weight to the semantic distances in the tentative arrangement order according to the cumulative number of times of the plurality of second keywords or/and the number of accumulated search times. An initial ranking order is generated by changing the order of the plurality of second keywords in the provisional arrangement order. For example, in the provisional arrangement order X ₀ ~X _n , the semantic distances of the second keyword X ₁ , X ₂ and the first keyword X ₀ are D(X ₀ , X ₁ )=0.3, D, respectively. X ₀ , X ₂ ) = 0.4, so X ₁ is in front of X ₂ . However, when the cumulative number of times of the second keyword X ₁ is counted (and/or the number of accumulated searched times) is 10, the cumulative number of times the second keyword X ₂ is counted (and/or the number of accumulated searched times) is 1000. If the threshold number of times preset by the search engine device is 100, and the first weight attached to the threshold of the number of times exceeds the semantic distance value -0.15, the value of D(X ₀ , X ₂ ) becomes 0.25, so that the second keyword The order of X ₁ and X ₂ is reversed, resulting in an initial ranking order. The above-mentioned cumulative number of clicks may be combined with the cumulative number of times of search, or may be calculated separately, and may be appropriately set by those skilled in the art. The above number of thresholds and first weights can be appropriately set by those skilled in the art. In addition, the number threshold may also be set to a plurality of values, and the plurality of thresholds may correspond to a plurality of first weight values. For example, when the number of times threshold exceeds 500, one of the first weight values is appended to -0.2, and when the number of times threshold exceeds 1000, one of the first weight values is appended to -0.25. If the semantic distance is less than zero after the weighting calculation, it is regarded as zero, and the semantic distance is no longer changed. If a plurality of second keywords are regarded as zero after the weighting calculation, the second keywords are arranged in the original order in the provisional arrangement order to generate an initial ranking order. The initial ranking order will be updated synchronously in order to generate the latest recommended list due to the increase in the number of times the cumulative number of clicks and the number of accumulated searches. In this way, the user can get a recommendation of the current hottest keyword from the recommendation list.

The communication module 40 receives the search string input from the user and transmits the search string to the search module 50. After receiving the search string, the search module 50 finds a first keyword that matches the search string and a plurality of second keywords that are defined by the affinity module as being determined according to the ranking module 35. The initial arrangement order displays a plurality of second keywords as a recommendation list, and the communication module 40 provides the user with further selection. When the search module 50 confirms that there is no first keyword that matches the search string, the search string and the accumulated searched number of times are stored in the database 10.

In addition to the above-described techniques for generating a recommendation list based on the cumulative number of times of being selected and the number of times of being searched, the creation of the recommendation list may be further generated based on the number of associations. Compared with the cumulative number of times of being selected and the number of times of being searched, the number of associations can avoid the problem that new keywords are not yet clicked, but are relatively large in number but not easily prioritized. That is to say, the keyword attribute data of the present creation further includes an associated quantity, which refers to the number of pieces of data related to a particular keyword, that is, the number of real estate objects that can be searched for when searching for the keyword.

When the search module 50 confirms that there is a first keyword that matches the search string, the search module 50 removes the second keyword whose number of associations is greater than a preset amount from the recommendation list to prevent a non-representative key. Words are recommended to reduce search efficiency. This is because the number of associations for a particular keyword is too large, which means that even if you search for this keyword, it will not be easy to find the real estate object you need because there is too much data. Again, the preset number can be suitably set by those skilled in the art. The sorting module 35 may also add a second weight to the semantic distance according to the number of associations, and generate an initial sorting order by changing the order of the plurality of second keywords in the tentative sorting order. For example, in the provisional arrangement order X ₀ ~X _n , the semantic distances of the second keyword X ₁ , X ₂ and the first keyword X ₀ are D(X ₀ , X ₁ )=0.2, D, respectively. X ₀ , X ₂ ) = 0.3, so X ₁ is in front of X ₂ . However, when the number of associations of the second keyword X ₁ is 300 and the number of associations of the second keyword X ₂ is 20, if the threshold number of associations preset by the search engine device is 100, the second number attached to the threshold number of associations is exceeded. The weight is a semantic distance value of 0.15, and the value of D(X ₀ , X ₁ ) is 0.35 after being weighted, so that the order of the second keywords X ₁ and X ₂ is reversed, and an initial ranking order is generated. Again, the associated quantity threshold and the second weight can be appropriately set by those skilled in the art.

In addition, when the search module 50 confirms that there is a first keyword that matches the search string, the plurality of second keywords may be included in the recommended list together with the corresponding plurality of associated numbers, and presented directly to the user. Help users decide to click. Please refer to the fifth figure, which is a schematic diagram of a recommended list of one embodiment of the present creation. In the fifth figure, each keyword in the recommendation list is accompanied by its corresponding associated quantity. Of course, in addition to the associated number, the sum of the number of times of the selected points in the attribute data, the number of accumulated searches, or the number of accumulated points selected and the number of accumulated searches may be presented.

In summary, the creation can record the attribute data selected by the user or the corresponding number of cumulative clicks, the cumulative number of searches and the number of related real estate objects, based on the real estate object data, by the attribute data size and The change identifies the popularity and representativeness of the keyword, and can add different weights to the keywords in the recommendation list to determine and instantly update the order of the keywords in the recommendation list to produce more accurate and convenient Recommend lists to improve search efficiency and help users find him as soon as possible without being familiar with real estate-related vocabulary The object you want to view (for example, unfamiliar with the South House is actually the Nangang National House).

The above is only the preferred embodiment of the present invention, and the scope of the present invention cannot be limited thereto. The simple equivalent changes and modifications made according to the scope of the patent application and the contents of the manual are still the present creation. Within the scope covered.

1‧‧‧Search engine unit

10‧‧‧Database

11‧‧ Real estate object database

12‧‧‧Real estate related vocabulary database

20‧‧‧Analysis module

30‧‧‧Affinity module

35‧‧‧Sorting module

40‧‧‧Communication module

50‧‧‧Search Module

Claims (10)

A search engine device capable of collecting keywords, the device comprising: a database configured to store a plurality of first keywords, and each of the first keywords having an attribute data, the attribute data including at least one accumulated The number of clicks and a cumulative number of searches; an analysis module configured to analyze a string in a source of data, and when the string is analyzed for real estate relevance, the string is defined as the number a keyword is stored in the database; an association module is configured to establish an association between the plural first keywords by using an inter-lexical correlation analysis technique, so that each first key The word is related to the plural second keyword, and is recorded in an association data, the plural second keyword is from the first keyword, and the attribute data possessed by the first keyword is used; a sorting module And configured to determine, according to the accumulated number of selected points of the plural second keywords or/and the accumulated number of searched times, one of each of the plural second keywords related to each first keyword is initially determined. Arranging a sequence; a communication module receiving a search string input from a user; and a search module configured to receive the search string and find the first keyword that matches the search string And the related module is defined as the second keyword corresponding to the plural number, and the second keyword is displayed as a recommendation list according to the initial ranking order determined by the sorting module. The search engine device of claim 1, wherein the database further comprises a real estate related vocabulary database, and the plurality of real estate related words are recorded, wherein the plurality of real estate related words Real estate related terms include: a county name, a city name, a township name, a town name, a district name, a li name, a street name, a well-known landmark name, a public facility name, a known community name, and A known object abbreviation is used as the analysis basis for the analysis module. The search engine device of claim 1, wherein the data source is a real estate object database or the communication module, and the real estate object database record includes at least one object description, one address, and a community name. And at least two of the real estate object materials in the group formed by the short name, the plurality of first keywords include at least one address, a community name or an object short name. The search engine device of claim 1, wherein the connection module calculates a plurality of semantic distances between the plurality of first keywords by a semantic analysis algorithm, according to the plurality of semantic distances Establishing an association between each first keyword and the plural second keyword. The search engine device of claim 4, wherein the sorting module sorts each of the first keywords and the plurality of second keywords into a tentative order according to the plurality of semantic distances. The search engine device of claim 5, wherein the sorting module adds a first weight according to the cumulative number of selected points of the plurality of second keywords or/and the size of the accumulated searched times The initial ranking order is generated by changing the order of the plurality of second keywords in the provisional arrangement order for each of the semantic distances. The search engine device of claim 1, wherein when the search module confirms that the first keyword does not match the search string, storing the search string and the accumulated searched number to the database. The search engine device of claim 1, wherein the search module further records a specific second keyword selected in the recommendation list, and records the cumulative selection of the specific second keyword. The number of times, so that the sorting module updates the initial sorting order. The search engine device of claim 1, wherein the attribute data further includes an associated quantity, the number of real estate objects that can be searched for when searching for the keyword. The search engine device of claim 9, wherein when the search module confirms that the first keyword matches the search string, the search module removes the number of associations from the recommended list by more than one a preset number of the second keywords; the sorting module adds a second weight to each of the semantic distances according to the size of the association, by changing the plurality of second keywords in the provisional arrangement order Sequence, the initial ranking order is generated; when the search module confirms that the first keyword matches the search string, the search module lists the plurality of second keywords together with the corresponding plurality of associated numbers Enter the list of recommendations.