CN119807343A - An intelligent question-answering system - Google Patents

Abstract

The invention discloses an intelligent question-answering system, which relates to the technical field of intelligent question-answering and comprises a question input module, an answer presentation module and a feedback collection module which are arranged on a front-end user interface, a question analysis module, an answer generation module and a knowledge base which are arranged on a back-end server, wherein the question input module is used for realizing the input of questions, the question analysis module can accurately analyze the meaning of the questions and identify the key and implicit requirements of the questions, the answer generation module is used for searching the knowledge base to generate answers based on the analysis results of the question analysis module, the answer presentation module is used for presenting the answers, the feedback collection module is used for collecting the evaluation and feedback of the answers of users and generating feedback information, and the knowledge base is used for storing the knowledge and the feedback information in the field. The invention improves the user experience, improves the question-answering efficiency, can promote the development of an intelligent question-answering system, and can adapt to diversified application scenes.

Description

Intelligent question-answering system

Technical Field

The invention relates to the technical field of intelligent question and answer, in particular to an intelligent question and answer system.

Background

The statements in this section merely provide background information related to the present disclosure and may not constitute prior art.

With the continuous development of artificial intelligence technology, intelligent question-answering systems are widely used in various fields. However, the different intelligent question-answering systems at present have great difference in the interaction flow, so that the user experience is uneven, and the question-answering efficiency is difficult to ensure. The lack of unified interactive flow specifications has limited the development of intelligent question-answering systems.

In the problem input link, the difference appears very obvious. Some intelligent question-answering systems only support traditional text entry, which undoubtedly increases the difficulty of operation for those users who are accustomed to using voice input. For example, in a busy working scenario, the user may be occupied with both hands and cannot enter text, and if the system does not support voice input, the efficiency of the user to acquire information is greatly affected. Still other systems, while supporting voice input, do not have high accuracy in recognizing voice, resulting in the need for repeated inputs by the user to allow the system to correctly understand the problem, wasting a lot of time and effort. In addition, the preliminary validity check criteria for the input content are also different. Some systems are too tightly checked, some somewhat special expressions may be judged as illegal, while some systems are too loosely checked, and are easy to accept some ambiguous and even erroneous inputs, which all bring great uncertainty to subsequent question analysis and answer generation.

Entering a problem analysis stage, wherein the natural language processing technology adopted by different intelligent question-answering systems is uneven. Some systems have major drawbacks in terms of semantic understanding, and for complex problem sentences, their true meaning cannot be resolved accurately. For example, when a user presents a question containing multiple clauses and modifiers, some systems may misunderstand the focus of the question and extract the wrong keywords. The algorithms and criteria vary from system to system in terms of keyword extraction. Some systems focus too much on the extraction of high frequency words, but ignore some low frequency words with specific meaning, resulting in a failure to fully grasp the user's needs. Some systems are easy to be interfered by noise in the keyword extraction process, and some vocabularies irrelevant to the problem are extracted, so that the accuracy of problem analysis is affected.

In the answer generation stage, the difference is also highlighted. Some intelligent question-answering systems cannot cover a wide knowledge field because the knowledge base is not perfect enough, and when users raise some questions of major or cold, the system cannot retrieve relevant information, so that accurate answers cannot be generated. Some systems have rich knowledge bases, but intelligent algorithms are not advanced enough, and the answers are not logical and rational in generating the answers, so that the answers are either too lengthy and complex to be understood by a user, or too short and fuzzy to meet the requirements of the user. In addition, there is a large gap between different systems in terms of accuracy and reliability of answers.

The uniqueness and non-uniformity of the form also presents an inconvenience to the user in terms of answer presentation. Some systems only adopt text form to present answers, and for some data or concepts needing visual presentation, effective expression cannot be carried out through forms such as charts. While some systems support multiple presentation forms, they are not convenient enough to switch and select, and users need to spend a lot of time searching for the appropriate presentation form. In addition, the answer presentation layout and typesetting of different systems are different, and the answer presentation layout and typesetting are too messy, so that the reading experience of a user is affected.

As for the feedback collection phase, the performances of different intelligent question-answering systems are quite different. Many systems either do not provide an explicit feedback channel and the user has no expression even if the answer is not satisfied. Or the importance of the feedback of the user is insufficient, and even if the feedback is received, the feedback cannot be processed and optimized in time. Some systems lack an effective incentive mechanism during feedback collection, and users have no incentive to provide feedback, thereby making the system incapable of continuous improvement.

The non-unification of the interaction flow ensures that users need to continuously adapt to new operation modes and experiences when using different intelligent question-answering systems, and the learning cost and the use difficulty of the users are increased. Meanwhile, due to the lack of unified standards, effective comparison and evaluation among all intelligent question-answering systems are difficult, and development of the whole industry is limited to a certain extent.

Disclosure of Invention

The invention aims to solve the problems in the prior art by providing an intelligent question-answering system.

The technical scheme of the invention is as follows:

The intelligent question-answering system comprises a question input module, an answer presentation module and a feedback collection module which are arranged on a front-end user interface, and a question analysis module, an answer generation module and a knowledge base which are arranged on a back-end server;

the system comprises a question input module, a question analysis module, an answer generation module, an answer presentation module and a feedback collection module, wherein the question input module is used for realizing the input of a question, the question analysis module can accurately analyze the meaning of the question and identify the key and implicit requirements of the question, the answer generation module is used for searching a knowledge base to generate an answer based on the analysis result of the question analysis module, the answer presentation module is used for presenting the answer, the feedback collection module is used for collecting the evaluation and feedback of the answer by a user and generating feedback information, and the knowledge base is used for storing the domain knowledge and the feedback information.

Further, the problem input module provides rich and various input modes so as to meet the requirements of different users in various scenes;

the problem input module is provided with a strict and intelligent validity checking mechanism for checking validity of the input problem.

Further, the input modes comprise text input, voice input and image input;

the method for verifying the validity of the input problem comprises the following steps:

for text input, not only grammar errors are checked, but also the rationality and the definiteness of the problem are judged, for voice input, advanced voice recognition technology is utilized to improve the accuracy, and at the same time, validity verification is carried out on the recognized text, for image input, key information is extracted through image recognition technology, and whether the key information meets the input requirement is judged.

Furthermore, the problem analysis module adopts advanced natural language processing technology and combines a deep learning algorithm to improve the accuracy of semantic understanding, and can accurately analyze the meaning of a complex problem statement and identify the key and implicit requirements of the problem.

Further, the problem analysis module accurately extracts keywords by analyzing sentence structures, semantic relationships and context information.

Furthermore, the problem analysis module is internally provided with a keyword extraction algorithm, comprehensively considers the frequency, importance and correlation of specific fields of the vocabularies, and simultaneously adopts a noise reduction technology to eliminate the interference of irrelevant vocabularies.

Further, the answer generation module is provided with an advanced intelligent algorithm, so that generated answers have logicality and regularity;

the answer generation module retrieves relevant information from the knowledge base based on the analysis result of the question analysis module and generates an answer.

Further, the answer presentation module provides answer presentation modes in various forms, including text, charts, voice and video, and automatically selects the most suitable presentation mode according to the type of the questions and the requirements of users.

Further, the feedback collection module is provided with a plurality of feedback channels, including online evaluation, opinion feedback forms and mail feedback;

the feedback collection module can also analyze and process user feedback in time based on a preset feedback processing mechanism, and optimize and improve the system according to feedback results.

Further, knowledge in a plurality of fields is covered in the knowledge base, and the knowledge base is updated and expanded continuously.

Compared with the prior art, the invention has the beneficial effects that:

1. The user experience is improved, through standardized interaction flow and rich and diverse functions, the user can more easily use the intelligent question-answering system to obtain accurate, timely and diverse answers, and the satisfaction degree of the user is greatly improved.

2. And the question-answering efficiency is improved, the standard interaction flow is beneficial to reducing unnecessary operation and waiting time, and the response speed of the intelligent question-answering system is improved. Meanwhile, the advanced algorithm and technology can rapidly and accurately analyze the questions and generate the answers, so that the question and answer efficiency is further improved.

3. The development of the intelligent question-answering system is promoted, the unified interactive flow specification provides common standards for different intelligent question-answering systems, and the development and innovation of the industry are promoted. The systems can continuously carry out technical upgrading and function expansion on the basis of conforming to specifications, and provide better services for users.

4. The method is suitable for diversified application scenes, and the standard interaction flow can ensure that intelligent question-answering systems in different fields have similar operation modes and user experience, so that a user can conveniently switch between different scenes. The unified interaction flow specification can enable the intelligent question-answering system to be better suitable for the characteristics of different devices and the habits of users, and consistent service experience is provided.

Drawings

FIG. 1 is a block diagram of an intelligent question-answering system;

fig. 2 is a flowchart of an intelligent question-answering standard interaction flow specification method.

Detailed Description

It is noted that relational terms such as "first" and "second", and the like, are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises an element.

The features and capabilities of the present invention are described in further detail below in connection with examples.

Example 1

Referring to fig. 1, an intelligent question-answering system specifically includes:

The system comprises a front-end user interface, a question input module, an answer presentation module, a feedback collection module, a question analysis module, an answer generation module and a knowledge base, wherein the question input module, the answer presentation module and the feedback collection module are arranged on the front-end user interface;

The system comprises a question input module, a question analysis module, an answer generation module, an answer presentation module, a feedback collection module and a knowledge base, wherein the question input module is used for realizing the input of a question, the question analysis module can accurately analyze the meaning of the question and identify the key and implicit requirement of the question;

The front-end user interface provides friendly and visual interaction environment, and is convenient for users to input questions and check answers. The interface design is succinct pleasing to the eye, the simple operation. Adopting a responsive design, adapting to the screen sizes and resolutions of different devices;

and the back-end server is responsible for the core tasks such as question analysis, answer generation, knowledge base management and the like. And the stability and the response speed of the system are ensured by adopting a high-performance server architecture and an advanced technology.

The system performance is continuously optimized, and the reliability and the expandability of the system are improved. The system is updated and upgraded regularly, and new technologies and functions are introduced to meet the changing demands of users.

In this embodiment, specifically, the problem input module provides rich and various input modes, so as to meet the requirements of different users in various scenes;

the problem input module is provided with a strict and intelligent validity checking mechanism for checking validity of the input problem.

In the embodiment, the specific input modes include text input, voice input and image input;

the method for verifying the validity of the input problem comprises the following steps:

for text input, not only grammar errors are checked, but also the rationality and the definiteness of the problem are judged, for voice input, advanced voice recognition technology is utilized to improve the accuracy, and at the same time, validity verification is carried out on the recognized text, for image input, key information is extracted through image recognition technology, and whether the key information meets the input requirement is judged.

In the embodiment, the problem analysis module adopts advanced natural language processing technology and combines a deep learning algorithm to improve the accuracy of semantic understanding, and can accurately analyze the meaning of a complex problem statement and identify the key and implicit requirements of the problem.

In this embodiment, specifically, the problem analysis module accurately extracts the keywords by analyzing the sentence structure, the semantic relationship, and the context information.

In this embodiment, specifically, the problem analysis module embeds a keyword extraction algorithm, comprehensively considers the frequency, importance and correlation of specific fields of the vocabulary, and simultaneously, the problem analysis module adopts a noise reduction technology to eliminate the interference of irrelevant vocabularies.

In this embodiment, specifically, the answer generation module has an advanced intelligent algorithm, so that the generated answer has logic and regularity;

the answer generation module retrieves relevant information from the knowledge base based on the analysis result of the question analysis module and generates an answer.

In this embodiment, the answer presentation module provides answer presentation modes in various forms, including text, chart, voice and video, and automatically selects the most suitable presentation mode according to the type of the question and the requirement of the user.

In this embodiment, the feedback collection module has multiple feedback channels, including online evaluation, opinion feedback form, and mail feedback;

the feedback collection module can also analyze and process user feedback in time based on a preset feedback processing mechanism, and optimize and improve the system according to feedback results.

In this embodiment, specifically, knowledge in multiple fields is covered in the knowledge base, and updating and expanding are continuously performed.

In this embodiment, it should also be noted that the above-mentioned intelligent question-answering system is designed and developed based on an intelligent question-answering standard interaction flow specification method.

In this embodiment, referring specifically to fig. 2, an intelligent question-answering standard interaction flow specification method includes:

step S1, designing an interaction flow of an intelligent question-answering system;

The interactive flow of the intelligent question-answering system is divided into five main stages of question input, question analysis, answer generation, answer presentation and feedback collection;

Wherein, the problem input stage:

Various input modes are provided, including text input, voice input, image input and the like, so as to meet the requirements of different users in various scenes. For example, a user may ask questions using voice input at the time of driving, and may ask questions related to a picture through image input at the time of viewing picture material;

specifically, for text input, advanced text input box technology is adopted, and functions such as automatic error correction and intelligent prompt are supported. For example, when a spelling error occurs during user input, the system automatically gives the correct spelling proposal. Meanwhile, an input length limit is set, so that the problem that the input of a user is overlong or too short is avoided;

for the aspect of voice input, a high-quality voice recognition engine is integrated, so that different accents and speech speeds can be adapted. When the user inputs voice, the system displays the recognition result in real time, so that the user can confirm conveniently. In addition, noise reduction processing is carried out on the input voice, so that the recognition accuracy is improved;

And for image input, analyzing the picture uploaded by the user by utilizing an image recognition technology, and extracting key information. For example, for pictures containing text, text is extracted by optical character recognition techniques and analyzed as part of the problem;

A strict and intelligent validity checking mechanism is established. For text entry, not only is grammar mistakes checked, but the question's rationality and clarity are also judged. For voice input, advanced voice recognition technology is utilized to improve accuracy, and at the same time, validity verification is carried out on the recognized text. For image input, extracting key information by an image recognition technology, and judging whether the key information meets the input requirement or not;

Problem analysis stage:

And deep understanding is carried out on the input problem by using the lexical analysis, the syntactic analysis and the semantic analysis technology in the natural language processing. And extracting keywords, topics and intentions in the questions to provide accurate basis for subsequent answer generation. Advanced natural language processing technology is adopted, and deep learning algorithm such as a large language model is combined, so that accuracy of semantic understanding is improved. For complex problem sentences, the meaning of the problem sentences can be accurately analyzed, and the key points and implicit requirements of the problems are identified. For example, keywords are accurately extracted by analyzing sentence structure, semantic relationships, and context information;

And establishing a problem classification system to classify different types of problems so as to adopt different processing strategies. For example, answers can be retrieved directly from a knowledge base for factual questions, and for inference-type questions, a logical inference algorithm is required to be applied for solving;

And optimizing a keyword extraction algorithm, and comprehensively considering the frequency, importance and relevance of a specific field of vocabulary. Not only high frequency vocabulary but also low frequency but critical vocabulary is not put through. Meanwhile, a noise reduction technology is adopted, so that interference of irrelevant words is eliminated, and the extracted keywords can accurately reflect the requirements of users;

answer generation stage:

The method comprises the steps of constructing a rich and comprehensive knowledge base, including structured data (such as table data in a database) and unstructured data (such as text files, webpage contents and the like), covering knowledge in multiple fields, and continuously updating and expanding. The knowledge map technology is adopted to correlate and integrate the knowledge in the knowledge base so as to more quickly retrieve the related information;

And according to the result of the question analysis, retrieving relevant knowledge segments from a knowledge base, and generating and optimizing answers by using an intelligent algorithm (such as a GPT series model). The generated answer is accurate, simple and easy to understand;

advanced intelligent algorithms are developed to enable logical and organized answers to be generated. The answer is neither too lengthy nor too vague, and the user questions can be answered accurately and succinctly. In the process of generating the answers, strict verification and screening are carried out, so that the accuracy and reliability of the answers are ensured;

answer presentation stage:

Various forms of answer presentation are provided, including text, graphics, speech, video, and the like. According to the type of the question and the requirement of the user, the most suitable presentation mode is automatically selected, for example, the answer can be directly presented in a text form for a simple question, a chart can be used for intuitively displaying a data type question, and a chart, an image or an animation video and the like can be used for displaying a complex question so as to enhance the intuitiveness and the readability of the answer;

Text presentation adopts clear and readable fonts and typesetting modes. For important content, thickening, coloring, etc. may be performed to attract the attention of the user. Meanwhile, the hyperlink and quotation functions are supported, so that the user can know related contents further;

the chart presents a chart form, such as a bar chart, a line chart, a pie chart, etc., that is selected appropriately according to the type of problem. Displaying complex data to a user in an intuitive form by utilizing a data visualization technology;

the voice presentation adopts a natural and smooth voice synthesis technology, and supports multiple languages and tone selection. The user can set the speech speed and volume of the speech according to the preference of the user;

and the layout and typesetting of the presentation interface are optimized, so that the answers are easy to read and understand. And meanwhile, related links and reference materials are provided, so that the user can conveniently and further know the background and related knowledge of the problem.

Feedback collection phase:

And establishing various feedback channels, such as online evaluation, opinion feedback forms, mail feedback and the like, so as to ensure that a user can conveniently evaluate and feed back answers, and providing a feedback button or an evaluation area after the answers are presented so as to encourage the user to evaluate and feed back the answers. The feedback content can comprise the accuracy, the completeness, the intelligibility and the like of the answer;

User feedback is highly valued, and a special feedback processing mechanism is established. The method comprises the steps of analyzing and processing user feedback in time, optimizing and improving a system according to feedback results, correcting and optimizing wrong answers in time, adding the wrong answers to a knowledge base, adding new questions raised by users to the knowledge base in time, and continuously perfecting an intelligent question-answering system;

and S2, establishing standardized specifications of the intelligent question-answering interaction flow.

In this embodiment, specifically, step S2 includes:

step S21, aiming at each stage of the interaction flow, a detailed and strict standardized specification is formulated;

the input format requirements of the problems include defining the format standards of different input modes, such as character limitation of text input, duration and tone quality requirements of voice input, resolution and format requirements of image input, etc.;

the problem analysis algorithm standard is that specific algorithms and technical indexes of semantic understanding and keyword extraction are specified, so that consistency and accuracy of different systems in the aspect of problem analysis are ensured;

the answer generation rule comprises the steps of formulating an answer generation principle and standard, wherein the answer generation principle and standard comprises the accuracy, the completeness and the logic requirement of an answer, the length and the language style specification of the answer and the like;

Determining specific format requirements of various presentation modes, such as fonts, word sizes and color specifications of texts, types and style standards of charts, speech speed and tone requirements of voices and the like;

Step S22, on the premise that different intelligent question-answering systems follow the specifications, consistency and high efficiency of the interaction flow can be achieved. Through the establishment of standardized specifications, the different systems can be effectively compared and evaluated, and the healthy development of the whole industry is promoted.

The above examples merely illustrate specific embodiments of the application, which are described in more detail and are not to be construed as limiting the scope of the application. It should be noted that it is possible for a person skilled in the art to make several variants and modifications without departing from the technical idea of the application, which fall within the scope of protection of the application.

This background section is provided to generally present the context of the present invention and the work of the presently named inventors, to the extent it is described in this background section, as well as the description of the present section as not otherwise qualify as prior art at the time of filing, are neither expressly nor impliedly admitted as prior art against the present invention.

Claims (10)

1. The intelligent question-answering system is characterized by comprising a question input module, an answer presentation module and a feedback collection module which are arranged on a front-end user interface, and a question analysis module, an answer generation module and a knowledge base which are arranged on a back-end server;

the system comprises a question input module, a question analysis module, an answer generation module, an answer presentation module and a feedback collection module, wherein the question input module is used for realizing the input of a question, the question analysis module can accurately analyze the meaning of the question and identify the key and implicit requirements of the question, the answer generation module is used for searching a knowledge base to generate an answer based on the analysis result of the question analysis module, the answer presentation module is used for presenting the answer, the feedback collection module is used for collecting the evaluation and feedback of the answer by a user and generating feedback information, and the knowledge base is used for storing the domain knowledge and the feedback information.

2. The intelligent question-answering system according to claim 1, wherein the question input module provides a rich and diverse input mode to meet the needs of different users in various scenarios;

the problem input module is provided with a strict and intelligent validity checking mechanism for checking validity of the input problem.

3. The intelligent question-answering system according to claim 2, wherein the input mode comprises text input, voice input, and image input;

the method for verifying the validity of the input problem comprises the following steps:

for text input, not only grammar errors are checked, but also the rationality and the definiteness of the problem are judged, for voice input, advanced voice recognition technology is utilized to improve the accuracy, and at the same time, validity verification is carried out on the recognized text, for image input, key information is extracted through image recognition technology, and whether the key information meets the input requirement is judged.

4. The intelligent question-answering system according to claim 3, wherein the question analysis module adopts advanced natural language processing technology and combines deep learning algorithm to improve accuracy of semantic understanding, and can accurately analyze meaning of complex question sentences and identify key and implicit demands of questions.

5. The intelligent question-answering system according to claim 4, wherein the question analysis module accurately extracts keywords by analyzing sentence structure, semantic relationships, and context information.

6. The intelligent question-answering system according to claim 5, wherein the question analysis module incorporates a keyword extraction algorithm to comprehensively consider the frequency, importance and relevance of specific fields of the vocabulary, and simultaneously, the question analysis module employs a noise reduction technique to eliminate the interference of irrelevant vocabularies.

7. The intelligent question-answering system according to claim 6, wherein the answer generation module has an advanced intelligent algorithm to provide logic and regularity to the generated answers;

the answer generation module retrieves relevant information from the knowledge base based on the analysis result of the question analysis module and generates an answer.

8. The intelligent question-answering system according to claim 7, wherein the answer presentation module provides a plurality of answer presentation modes including text, chart, voice and video, and automatically selects the most suitable presentation mode according to the type of the question and the requirement of the user.

9. The intelligent question-answering system according to claim 8, wherein the feedback collection module has a plurality of feedback channels, including on-line evaluation, opinion feedback forms, mail feedback;

the feedback collection module can also analyze and process user feedback in time based on a preset feedback processing mechanism, and optimize and improve the system according to feedback results.

10. The intelligent question-answering system according to claim 9, wherein knowledge of multiple domains is covered in the knowledge base and is updated and expanded continuously.