Intelligent question answering system and method based on knowledge base

Abstract

The invention relates to the technical field of data processing, in particular to an intelligent question answering system and method based on a knowledge base, and discloses an intelligent question answering method based on the knowledge base, which comprises the following steps: establishing a triple data set; the method comprises the following steps: performing semantic analysis on a question proposed by a user through a semantic analysis module, extracting a question focus in the question, and converting the question focus into a triple data form without an actual answer; and linking the marker of each data in the triple according to the orientation of the address symbol to form a data chain, comparing the triple data form without the actual answer with the data chain of the approximate triple data in the triple data set, screening out the data with the highest similarity for integration, and outputting the integration result as the answer of the question. According to the method, the data are integrated in a triple data mode to form the triple data set, and the triple data form without actual answers is formed in a simulated answer mode, so that answers contained in a knowledge base can be quickly found, and the question and answer mode reduces the dependence on semantic analysis results.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an intelligent question-answering system and method based on a knowledge base. Background Art

[0002] With the continuous development of technology, people's technical solutions for data processing have become increasingly simplified, from the initial written records to data storage on readable media, from the initial complex classifications to today's classified databases. In order to facilitate data queries, various data retrieval systems have been designed and applied. Taking the intelligent customer service of major application platforms as an example, automatic replies are provided to users by querying relevant knowledge bases. The control of these automatic replies is usually carried out by semantically analyzing the questions raised by users, and analyzing and retrieving the relevant knowledge in the knowledge base based on the results of the semantic analysis. Finally, the result closest to the question raised by the user is obtained, and this result is output to the user as the answer. This analysis method is overly dependent on the accuracy of the semantic analysis results. If a semantic analysis error occurs, the output result will not match the actual problem. Summary of the invention

[0003] The present invention aims at the deficiencies in the prior art and provides the following technical solutions: An intelligent question answering method based on a knowledge base comprises the following steps: S10: Establishing a triple data set, wherein each unit of the triple data stores an address symbol and a marker in addition to the text data, and the address symbol points to another data in the triple data in a semantic order, and the marker marks the data with the same semantics as equivalent; S20: performing semantic analysis on the question raised by the user through a semantic analysis module, extracting the focus of the question, and converting the focus of the question into a triple data form without an actual answer; S30: Link the identifier of each data in the triple with the direction of the address symbol to form a data chain, compare the triple data form without actual answer with the data chain of approximate triple data in the triple data set, select the data with the highest similarity for integration, and output the integration result as the answer to the question.

[0004] As an improvement of the above technical solution, the problem focus includes entity features in the problem and relationship expressions between entity features.

[0005] As an improvement of the above technical solution, the method for extracting the focus of the question in step S20 includes the following steps: S21: Output intent category of the question identified by BERT fine-tuning model.

[0006] S22: The entity features in the question are extracted through the BiLSTM-CRF model, and the entity features are linked to the triple data set. If the entity features do not exist, it is marked as an unverified entity. The entity features, the relationship between the entity features, and the unverified entities are output in the form of triple data to form a triple data form without an actual answer.

[0007] As an improvement of the above technical solution, step S30 includes the following steps: S31: List all triplet data in the triplet data set that are identical to the entity feature characters extracted in step S22.

[0008] S32: Compare the entity features extracted from the question with the data in the triple data to obtain several groups of triple data with the highest similarity.

[0009] S33: Obtain approximate triple data from the triple data, extract the data with the same position as the unverified entity for integration, and output the integration result as an answer.

[0010] As an improvement of the above technical solution, the method for obtaining several groups of triplet data with the highest similarity in step S32 depends on the following steps: S321: Compare the entity features extracted from the question with the triple data, extract all triple data with the entity features as the main body, and classify them according to the markers of the entity features to obtain triple data with the same characters and the same meaning.

[0011] S322: Filter out several groups of triplet data with the highest similarity to the meaning of the relational expression from the acquired triplet data and output them.

[0012] As an improvement of the above technical solution, the acquisition of the similarity of the meaning of the relationship expression in step S322 depends on the following steps: S3221: Preset a similarity threshold. When the calculated similarity is greater than the threshold, the words are considered to have the same meaning.

[0013] S3222: Obtain the vocabulary in the triple data with the same position as the relationship expression through the semantic analysis module, and obtain the similarity between the vocabulary and the vocabulary expression in the triple data form without actual answers, and filter out the data with similarity greater than a preset threshold as the output of several groups of triple data with the highest similarity.

[0014] As an improvement of the above technical solution, the data chain in step S33 includes a comparison chain and a pointing chain, and step S33 includes the following steps: S331: Obtain the address symbol pointing of the entity feature in the triple data form without actual answer, and combine the markers of the entity feature in the triple data in the direction pointed by its address symbol to form a comparison chain.

[0015] S332: Obtain the directions of the address symbols of the entity features in the plurality of triple data with the highest similarity, and combine the markers of the entity features in each triple data in the direction pointed by their address symbols to form a pointing chain.

[0016] S333: Compare the pointing chain with the comparison chain. If the directions are completely the same or completely opposite, it is determined to be approximate triplet data.

[0017] S334: Integrate the data in the approximate triple data that are in the same position as the unverified entity and replace the content in the unverified entity to obtain triple data with an actual answer, and convert the triple data into a complete sentence and output it as an answer.

[0018] A knowledge base-based intelligent question-answering system operates using a knowledge base-based intelligent question-answering method as described in the aforementioned technical solution, including: a database, a semantic analysis module, a data processing module, and a data interaction module.

[0019] The database stores at least triple data sets and corpus data. The semantic analysis module is used to perform semantic analysis on questions raised by users, calculate the similarity of words, and determine whether the meanings of words are the same based on the similarity. The data processing module is used to extract the focus of the question and convert the focus of the question into a triple data form without an actual answer. The data interaction module is used to interact with users on data.

[0020] As an improvement of the above technical solution, the database further includes a marker table, and the marker table stores all markers and the vocabulary corresponding to the markers.

[0021] Beneficial effects of the present invention: A triple data set is formed by integrating the data in the form of triple data, and a triple data form without actual answers is formed in the form of simulated answers. By comparing and screening the triple data form without actual answers with the integrated triple data set, the answers contained in the knowledge base can be quickly found. This question-and-answer method relies more on the establishment of a triple data set. As long as the accuracy can be guaranteed in the early stage of the establishment of the triple data set, the accuracy of subsequent analysis results can be guaranteed, reducing the dependence on semantic analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A flowchart of an intelligent question-answering method based on a knowledge base of the present invention; Figure 2 is a flowchart of step S20 in a knowledge base-based intelligent question answering method of the present invention; Figure 3 It is a flowchart of step S30 in the knowledge base-based intelligent question answering method of the present invention. DETAILED DESCRIPTION

[0023] The following describes the embodiments of the present invention through specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention.

[0024] In actual applications, the intelligent customer service of major application platforms usually provides automatic replies to users by querying relevant knowledge bases. These automatic replies are usually controlled by performing semantic analysis on the questions raised by users, and analyzing and retrieving the relevant knowledge in the knowledge base based on the results of the semantic analysis. Finally, the result closest to the question raised by the user is obtained, and this result is output to the user as the answer. This analysis method is overly dependent on the accuracy of the semantic analysis results. If a semantic analysis error occurs, the output result will be inconsistent with the actual problem.

[0025] In order to solve the above problems, the following embodiments are provided: Embodiment 1 See also Figures 1 to 3 , provides an intelligent question answering method based on a knowledge base, comprising the following steps: S10: Establish a triple data set, wherein each unit of the triple data stores an address symbol and a marker in addition to text data, and the address symbol points to another data in the triple data in a semantic order, and the marker marks the data with the same semantics as equivalent.

[0026] When establishing a triple data set, it is necessary to consider the way in which the triple data set is established. Since the existing technology leads to uncertainty in the results based on semantic judgment, in order to reduce this uncertainty, the data needs to be integrated in a certain way. Therefore, it is necessary to ensure the unique pointing of the data. Initial consideration is given to adding address symbols for determination. Since various types of text have non-unique meanings, such as various polyphones and polysemous words, it is also necessary to ensure that when a unique character is located, the word can be accurately classified. Here, consider adding markers and marking the markers with the semantics of the characters. The semantic marking here can be machine judgment or human judgment, or a combination of the two. Since it is the initial establishment of the data set, this establishment is controllable and can be adjusted later. Therefore, although it is also judged by semantics, it has a high degree of adjustability and can be adjusted later, which can minimize the uncertainty brought by semantic analysis.

[0027] Take "Xiao Ming is Xiao Gang's father" as an example. When the sentence is used to create triple data, the extracted entity features are "Xiao Ming" and "Xiao Gang", and "father" is stored as a relational expression, that is, a triple data of ["Xiao Ming", "father", "Xiao Gang"] is obtained. When address symbols and marker codes are added, "father" has an address symbol pointing to "Xiao Ming", and "Xiao Ming" has an address symbol pointing to "Xiao Gang". As a child, "Xiao Gang" will establish an address symbol pointing to "father". The address symbols here can form a chain with their pointing, and use this chain to form multiple sentences with different orders or characters but the same meaning, for example: "Xiao Ming is Xiao Gang's father", "Xiao Gang's father is Xiao Ming", and "The person whose father is Xiao Ming is Xiao Gang".

[0028] When marker codes are added, the different meanings of words can be distinguished. Take "apple" as an example. When it involves words such as "performance" and "parameters", its meaning refers to Apple or Apple mobile phones. When it involves words such as "taste" and "planting", its meaning refers to a type of fruit. Different marker codes indicate different meanings for the same characters. As long as the marker code is determined, all triple data with direct adjacent relationships under this marker code can be determined.

[0029] After the establishment of the triple data set, it is necessary to analyze the user's question. Different from the traditional semantic analysis, after the triple data set is established, it is not necessary to completely determine the specific semantic situation, but only to find the focus of the question raised by the user as much as possible. Specifically, step S20 is executed.

[0030] S20: Perform semantic analysis on the question raised by the user through the semantic analysis module, extract the focus of the question, and convert the focus of the question into a triple data form without an actual answer.

[0031] The problem focus includes entity features in the problem and the relationship between entity features. Based on this, the method for extracting the problem focus in the problem in step S20 includes the following steps: S21: Output intent category of the question identified by BERT fine-tuning model.

[0032] First of all, we need to know the intention of the question raised by the user, whether it is to find a practical answer or a hypothetical question. For example, "What is the speed of sound propagation in a vacuum?" Since there is no medium in a vacuum, sound cannot propagate in a vacuum environment. Although this question does not have an answer in our conventional semantic understanding, it should be classified as a question that seeks a practical answer. For example, "If Napoleon had won the Battle of Waterloo, what would Europe be like now?" This question should be a hypothetical question. For this kind of hypothetical question, in this embodiment, usually only "cannot be analyzed based on existing knowledge" is output. This is because the complexity of finding answers to hypothetical questions is too high. The design of this embodiment is to speed up the query of data. Therefore, the analysis process of this type of hypothetical question is deleted to speed up the data analysis. When applied to large-scale data retrieval, this processing method can greatly speed up the analysis and retrieval of data. Of course, users can also process this kind of hypothetical question by embedding an independent functional module and provide an answer mode for the hypothetical question. However, in this embodiment, only questions that seek practical answers are considered, and subsequent functions are started for the question.

[0033] When step S21 determines that the user's question is a question seeking a practical answer, step S22 is executed.

[0034] S22: The entity features in the question are extracted through the BiLSTM-CRF model, and the entity features are linked to the triple data set. If the entity features do not exist, it is marked as an unverified entity. The entity features, the relationship between the entity features, and the unverified entities are output in the form of triple data to form a triple data form without an actual answer.

[0035] In this embodiment, the BERT fine-tuning model and the BiLSTM-CRF model are both recognition models recorded in the prior art, wherein the BERT fine-tuning model is a pre-trained natural language processing model that can consider the vocabulary of the context and better understand the meaning of the sentence. It can distinguish the type of user questions and determine the output intent category based on the content of the user's questions, while the BiLSTM-CRF model processes sequence labeling tasks in natural language processing. Its main body is composed of a bidirectional long short-term memory network (Bi-LSTM) and a conditional random field (CRF). The model input is character features, and the output is the predicted label corresponding to each character. By labeling the vocabulary in the question, the function of extracting entity features is realized (entity features are usually the subject words in a sentence. There is at least one subject word in a sentence, and there may be multiple subject words).

[0036] Similar to the establishment of the previous triple data set, the extraction here prioritizes the extraction of entity features. For example, "Who invented the camera?" When performing entity feature extraction, two entities "camera" and "inventor" can be extracted, and the relationship between these two entity features, "invention", can be obtained. However, there is no actual entity in the inventor column, that is, the entity feature does not exist. Here we mark it as an unverified entity. When we answer, the result should be "The camera was invented by ***" or "The inventor of the camera is ***", and this answer form is converted into a triple form: ["Unknown inventor", "invention", "camera"], which is a triple data form without an actual answer.

[0037] After obtaining the triple data form, comparison and screening need to be performed, and step S30 is continued here.

[0038] S30: Link the identifier of each data in the triple with the direction of the address symbol to form a data chain, compare the triple data form without actual answer with the data chain of approximate triple data in the triple data set, select the data with the highest similarity for integration, and output the integration result as the answer to the question.

[0039] Specifically, step S30 includes the following steps: S31: List all triplet data in the triplet data set that are identical to the entity feature characters extracted in step S22.

[0040] Regardless of other features, the entity feature should always exist in the question. This entity feature usually cannot be changed by using similar words. Therefore, you only need to filter out the triple data with the same characters as the entity feature in the triple data set to obtain the data with the correct answer.

[0041] In the aforementioned extraction of entity features, the relationship expression between entity features is usually considered to be extractable, but it cannot be obtained in some cases, such as "What is the relationship between Xiao Ming and Xiao Gang". In this case, only the entity features "Xiao Ming" and "Xiao Gang" can be extracted, and an unknown relationship expression can be obtained. When establishing a triple data form without an actual answer, it is: ["Xiao Ming", "Unknown relationship", "Xiao Gang"]. However, if this data exists in the database, the entity features "Xiao Ming" and "Xiao Gang" are unique, and the relationship expression can be reversed through these two entity features. Therefore, it is only necessary to determine that the characters of the entity features are the same.

[0042] S32: Compare the entity features extracted from the question with the data in the triple data to obtain several groups of triple data with the highest similarity; After the extraction of triple data is completed, it is necessary to obtain triple data with the highest similarity from step S32, in order to further reduce the complexity of data comparison and further screen the data of the triple data set, specifically, including the following steps: S321: Compare the entity features extracted from the question with the triple data, extract all triple data with the entity features as the main body, and classify them according to the markers of the entity features to obtain triple data with the same characters and the same meaning.

[0043] Since the meaning represented by the marker is unique, we only need to filter out the corresponding triple data when the meaning is the same, which can ensure that the filtered triple data has the characteristics of positive correlation. For example, the question is "What tools did Xiao Ming invent?", which is converted into triple data without actual answers: ["Xiao Ming", "invention", "unknown tool"]. When the entity feature "Xiao Ming" is used as the subject, since the marker of "Xiao Ming" is unique and the meaning is the same, after filtering with the entity feature "Xiao Ming" as the subject, all the triple data obtained are positively correlated with "Xiao Ming". After the data is converted into regular sentences, it may be "Xiao Ming is Xiao Gang's father", "Xiao Ming invented the hammer", etc.

[0044] After the positively associated data is determined, further differentiation is performed through relationship expression, specifically, step S322 is performed.

[0045] S322: Filter out several groups of triplet data with the highest similarity to the meaning of the relational expression from the acquired triplet data and output them.

[0046] Specifically, the acquisition of the similarity of the meaning of the relationship expression in step S322 depends on the following steps: S3221: Preset a similarity threshold. When the calculated similarity is greater than the threshold, the words are considered to have the same meaning.

[0047] Different from the semantic analysis of traditional whole sentence questions, in this embodiment, only a small amount of word meanings need to be analyzed to make a judgment, and when errors occur in the results, only the corresponding threshold needs to be increased.

[0048] S3222: Obtain the vocabulary in the triple data with the same position as the relationship expression through the semantic analysis module, and obtain the similarity between the vocabulary and the vocabulary expression in the triple data form without actual answers, and filter out the data with similarity greater than a preset threshold as the output of several groups of triple data with the highest similarity.

[0049] As mentioned in the establishment of the aforementioned triple data set, each unit in each triple data has an address symbol and a marker symbol. Therefore, as long as two data in the triple data are located, the position of the third data in the triple data can be known. When the meaning of the word expression in the triple data form without an actual answer is similar, it is considered to be highly similar and may be synonymous or the same type of words. For example, "Xiao Ming is Xiao Gang's father", "Xiao Gang is Xiao Ming's son", and "Xiao Gang injured Xiao Ming". When these three sentences are converted into triple form, the results are: ["Xiao Ming", "father", "Xiao Gang"], ["Xiao Gang", "son", "Xiao Ming"], ["Xiao Ming", "wound", "Xiao Gang"], in this context, it can be considered that "father" and "son" are words that express the relationship between characters, and "wound" is an action verb. During analysis, the action verb "wound" is different from the previous two parts of speech. When the entity feature "Xiao Ming" is exactly the same as the entity feature "Xiao Gang", it can be considered that the relationship expression words "son" and "father" are expressions with a high degree of similarity. Therefore, when outputting, this type of vocabulary will also be output together.

[0050] S33: Obtain approximate triple data from the triple data, extract the data with the same position as the unverified entity for integration, and output the integration result as an answer.

[0051] As analyzed above, since the triple data without actual answers are also obtained according to the establishment method of the initial triple data set, if the approximate triple data has an actual answer, then the triple data without actual answers and the approximate triple data must have a group of address symbols pointing to completely the same or completely opposite data. Specifically, the data chain in step S33 includes a comparison chain and a pointing chain, and step S33 includes the following steps: S331: Obtain the address symbol pointing of the entity feature in the triple data form without actual answer, and combine the markers of the entity feature in the triple data in the direction pointed by its address symbol to form a comparison chain.

[0052] S332: Obtain the directions of the address symbols of the entity features in the plurality of triple data with the highest similarity, and combine the markers of the entity features in each triple data in the direction pointed by their address symbols to form a pointing chain.

[0053] S333: Compare the pointing chain with the comparison chain. If the directions are completely the same or completely opposite, it is determined to be approximate triplet data.

[0054] For example, in the two sentences "Xiao Ming is Xiao Gang's father" and "Xiao Ming's son is Xiao Gang", when the address symbols are analyzed in the above way, the first sentence will get the address symbol of "father" pointing to "Xiao Ming", the address symbol of "Xiao Ming" pointing to "Xiao Gang", and the address symbol of "Xiao Gang" pointing to "father". The second sentence will get the address symbol of "son" pointing to "Xiao Gang", the address symbol of "Xiao Gang" pointing to "Xiao Ming", and the address symbol of "Xiao Ming" pointing to "son". Convert it into the form of mark code. If the expression form of mark code is added, "Xiao Ming" is represented by 1, "father" and "son" are expressed as 2 as the relationship between characters, and "Xiao Gang" is represented by 3. The resulting pointing chain is as well as , because in this embodiment, the triple data is established based on the relational expression, therefore, the tag code of the relational expression can be ignored here, and only the tag code of the entity feature can be considered, that is, the resulting pointing chain should be as well as , and the directions of these two pointing chains are usually the same or opposite as long as the answers are consistent in the above steps.

[0055] In addition to the above situation, there are two families with both "Xiao Ming" and "Xiao Gang", but the relationship between the two is different, for example, "Xiao Ming is Xiao Gang's uncle". This situation seems to lead to confusing output results, but from the establishment of the initial data set, we can know that the establishment of the marker code not only refers to a single character, but also contains meaning. When the family is taken as the unit, the meaning of the marker code is the family relationship chain. Unless the names of all members of the two families are exactly the same, there are differences in the family relationship chain. When generating marker codes, different marker codes will be generated. This probability can be ignored, so it can be ignored.

[0056] S334: Integrate the data in the approximate triple data that are in the same position as the unverified entity and replace the content in the unverified entity to obtain triple data with an actual answer, and convert the triple data into a complete sentence and output it as an answer.

[0057] If there are multiple approximate triple data, the answer is considered to be mergeable. For example, the question "What tool did Xiao Ming invent?" can have multiple answers, that is, Xiao Ming invented more than one tool, so there will be multiple approximate triple data in the final result. At this time, all the data with the same position as the unverified entity can be integrated, and then the content in the unverified entity can be replaced to obtain the triple data with the actual answer. Unlike conventional triple data, the data filled in the position to replace the unverified entity is in array form. When outputting, the data in the array is traversed and integrated with the entity features and relationship expressions to form a complete sentence as the answer for output. This integrated data can be stored in the triple data, which can speed up the query analysis when encountering corresponding questions later.

[0058] In this embodiment, unknown questions are not analyzed. This is because this embodiment is mostly used for question and answer in a specific knowledge base. The questioner should ask questions based on the corresponding knowledge system. This way of asking questions may have incomplete answers, but there will usually be certain answers. If there is no accurate answer to the question asked, in this embodiment, only "No accurate result was found" will be output, and similar questions with answers will be provided for users to choose from, so as to avoid query errors caused by user input errors.

[0059] Embodiment 2 Provided is a knowledge base-based intelligent question answering system, which is operated using a knowledge base-based intelligent question answering method as described in Example 2, and includes: a database, a semantic analysis module, a data processing module, and a data interaction module.

[0060] The database stores at least triple data sets and corpus data. The semantic analysis module is used to perform semantic analysis on questions raised by users, calculate the similarity of words, and determine whether the meanings of words are the same based on the similarity. The data processing module is used to extract the focus of the question and convert the focus of the question into a triple data form without an actual answer. The data interaction module is used to interact with users on data.

[0061] The semantic analysis module obtains the questions raised by the user, performs word meaning analysis and calculates vocabulary similarity to determine whether the vocabulary meanings are the same or similar, and extracts the focus of the question through the data processing module, and converts the focus of the question into a triple data form without an actual answer. After comparing and processing the similarity, the analysis results are output to the user through the data processing module.

[0062] In one embodiment, the database further includes a tag table, and the tag table stores all tags and the vocabulary corresponding to the tags.

[0063] In the above scheme, the markers can also be recorded separately in a table, so that the query of the markers is more efficient and faster.

[0064] The above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them. Anyone familiar with the technology can modify or change the above embodiments without violating the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by those with ordinary knowledge in the technical field without departing from the spirit and technical ideas disclosed by the present invention should still be covered by the claims of the present invention.

Claims

1. An intelligent question answering method based on a knowledge base, characterized in that: The steps include: S10: Establishing a triple data set, wherein each unit of the triple data stores an address symbol and a marker in addition to the text data, and the address symbol points to another data in the triple data in a semantic order, and the marker marks the data with the same semantics as equivalent; S20: performing semantic analysis on the question raised by the user through a semantic analysis module, extracting the focus of the question, and converting the focus of the question into a triple data form without an actual answer; S30: Link the identifier of each data in the triple data with the direction of the address symbol to form a data chain, compare the triple data form without actual answer with the data chain of approximate triple data in the triple data set, select the data with the highest similarity for integration, and output the integration result as the answer to the question.

2. The intelligent question-answering method based on a knowledge base according to claim 1, characterized in that: The focus of the problem includes entity features in the problem and the relationship expression between entity features.

3. The intelligent question-answering method based on a knowledge base according to claim 2, characterized in that: The method for extracting the focus of the question in step S20 comprises the following steps: S21: Identify the output intent category of the question through the BERT fine-tuning model; S22: The entity features in the question are extracted through the BiLSTM-CRF model, and the entity features are linked to the triple data set. If the entity features do not exist, it is marked as an unverified entity. The entity features, the relationship between the entity features, and the unverified entities are output in the form of triple data to form a triple data form without an actual answer.

4. The intelligent question-answering method based on a knowledge base according to claim 3, characterized in that: The step S30 includes the following steps: S31: List all triplet data in the triplet data set that are identical to the entity feature characters extracted in step S22; S32: Compare the entity features extracted from the question with the data in the triple data to obtain several groups of triple data with the highest similarity; S33: Obtain approximate triple data from the triple data, extract the data with the same position as the unverified entity for integration, and output the integration result as an answer.

5. The intelligent question-answering method based on a knowledge base according to claim 4, characterized in that: The method for obtaining the highest similarity groups of triplet data in step S32 depends on the following steps: S321: Compare the entity features extracted from the question with the triple data, extract all triple data with the entity features as the main body, and classify them according to the markers of the entity features to obtain triple data with the same characters and the same meaning; S322: Filter out several groups of triplet data with the highest similarity to the meaning of the relational expression from the acquired triplet data and output them.

6. The intelligent question-answering method based on a knowledge base according to claim 5, characterized in that: The acquisition of the similarity of the meaning of the relationship expression in step S322 depends on the following steps: S3221: Preset a similarity threshold. When the calculated similarity is greater than the threshold, the words are considered to have the same meaning. S3222: Obtain the vocabulary in the triple data with the same position as the relationship expression through the semantic analysis module, and obtain the similarity between the vocabulary and the vocabulary expression in the triple data form without actual answers, and filter out the data with similarity greater than a preset threshold as the output of several groups of triple data with the highest similarity.

7. The intelligent question-answering method based on a knowledge base according to claim 4, characterized in that: The data chain in step S33 includes a comparison chain and a pointing chain, and step S33 includes the following steps: S331: obtaining the address symbol pointing of the entity feature in the triple data form without actual answer, and combining the markers of the entity feature in the triple data in the direction pointed by the address symbol to form a comparison chain; S332: Obtain the directions of the address symbols of the entity feature in the plurality of triple data with the highest similarity, and combine the markers of the entity feature in each triple data in the direction pointed by its address symbol to form a pointing chain; S333: Compare the pointing chain with the comparison chain, and if the directions are completely the same or completely opposite, it is determined to be approximate triplet data; S334: Integrate the data in the approximate triple data that are in the same position as the unverified entity and replace the content in the unverified entity to obtain triple data with an actual answer, and convert the triple data into a complete sentence and output it as an answer.

8. An intelligent question-answering system based on a knowledge base, operating using an intelligent question-answering method based on a knowledge base as claimed in any one of claims 1 to 7, characterized in that: include: A database stores at least a triple data set and corpus data; The semantic analysis module is used to perform semantic analysis on the questions raised by the user, calculate the similarity of the words, and determine whether the meanings of the words are the same based on the similarity; A data processing module is used to extract the focus of the question and convert the focus into a triple data form without an actual answer; The data interaction module is used to interact with users.

9. The intelligent question-answering system based on a knowledge base according to claim 8, characterized in that: The database also includes a marker table, which stores all markers and the vocabulary corresponding to the markers.

Images