CN116644756A - Semantic Analysis Method Based on Knowledge Graph - Google Patents

Abstract

The invention relates to the technical field of semantic analysis, in particular to a semantic analysis method based on a knowledge graph, which comprises the steps of obtaining voice data, screening the voice data, removing disordered information, leaving an effective sentence pattern, analyzing the effective sentence pattern, obtaining text information and tone information, identifying the text information and the tone information according to a pre-trained semantic model, analyzing the text information and the tone information through standard synonyms and a spoken word stock, obtaining first semantic information, carrying out association analysis according to the knowledge graph constructed by combining the first semantic information, obtaining second semantic information, comparing the second semantic information with the first semantic information, outputting the second semantic information when the difference value of a comparison result is lower than a set value, and reminding a user to reenter the difference value of the comparison result when the difference value of the comparison result is higher, so that semantic information which meets the requirements of a user can be obtained through comparison of double voice information, and the accuracy of the information in the semantic data is improved.

Description

Semantic Analysis Method Based on Knowledge Graph

technical field

The invention relates to the technical field of semantic analysis, in particular to a semantic analysis method based on a knowledge graph.

Background technique

In recent years, the development of speech recognition technology has greatly improved the level of human-computer interaction, and semantic analysis technology, as a key part of understanding natural language, has played a decisive role in the intelligence of human-computer interaction. However, from the perspective of natural language, most words have the characteristics of polysemy. In addition to expressing the original meaning, a word may also have other hidden semantics, which cannot be accurately identified only by keywords. its practical significance. In addition, when the sentence is a colloquial sentence, the predicate in the sentence may not be found during the semantic analysis process, making it difficult to achieve accurate semantic analysis;

The knowledge map is a combination of theories and methods of applied mathematics, graphics, information visualization technology, information science and other disciplines with metrology citation analysis, co-occurrence analysis and other methods, and uses the visual map to vividly display the core structure and development of the subject. History, frontier fields, and the overall knowledge structure achieve the modern theory of multidisciplinary integration. It displays complex knowledge fields through data mining, information processing, knowledge measurement and graphic drawing, reveals the dynamic development law of knowledge fields, and provides practical and valuable references for subject research.

Therefore, it is necessary to provide a semantic analysis method based on knowledge graphs to solve the above technical problems.

Contents of the invention

In order to solve the above technical problems, the present invention provides a semantic analysis method based on knowledge graph.

The semantic analysis method based on the knowledge map provided by the present invention, the specific steps are as follows:

S1. Acquire voice data, and screen the voice data, so as to eliminate messy information, leaving only effective sentence patterns, and analyze the effective sentence patterns to obtain text information and tone information, and combine the obtained text information and tone information input into the speech model;

S2. Recognize the text information and tone information according to the pre-trained semantic model, and analyze them through the standard synonyms and colloquial word lexicon to obtain the first semantic information;

S3. Perform association analysis based on the first semantic information combined with the constructed knowledge map to obtain second semantic information;

S4. Comparing the second semantic information with the first semantic information, when the difference value of the comparison result is lower than the set value, determine the semantic analysis result according to the second semantic information, and output it, and compare the difference value of the result When it is higher, the user is reminded to re-enter.

Preferably, the specific method of constructing the knowledge map is:

S31. Connect various data sources in a visual way, verify and manage the data sources, and establish a basic knowledge base according to the data sources;

S32. Analyze the data source, and generate multiple semantic vectors by analyzing it;

S33. Perform data fusion on the generated semantic vectors, and mine association relationships between the fused semantic vectors to generate relationships between semantic vectors;

S34, forming a semantic database from the fused semantic vector and the relationship between the semantic vectors, storing it in the server, evaluating the quality of the semantic database, and completing the construction of the knowledge map.

Preferably, the data sources in step S31 include standard synonyms and colloquial word thesaurus, basic information filled in by the user, user's usual behavior data, user's chat information and webpage data published on the Internet.

Preferably, the voice data is obtained, and the voice data is screened to remove the messy information, leaving only effective sentence patterns, and analyzing the effective sentence patterns to obtain the text information. The steps are:

S11. Extracting sound features from the speech data;

S12. Match the sound feature quantity with the modeled sound data in the sound library, so as to obtain sound data with similarity matching;

S13. Compare and match the voice data with the voice data stored in the text and voice database to obtain text information.

Preferably, the text and speech database stores text, speech corresponding to the text, and speech data of extended words and sentences of the text.

Preferably, the voice data is obtained, and the voice data is screened to remove the messy information, leaving only effective sentence patterns, and analyzing the effective sentence patterns to obtain the tone information as follows:

S14. Perform spectrum analysis on the voice data, and extract tone phonemes from the voice data;

S15. Match the tone of the voice data in the tone model according to the tone phoneme.

Preferably, the text information and tone information are identified according to the pre-trained semantic model, and analyzed through a standard synonym and colloquial word lexicon to obtain the first semantic information:

S21. Identify the tone information through the voice model, and obtain the tone order, and eliminate the combination of the tone order in the text information and the recognized tone order tone information through the tone information;

S22. Segment the text information through a semantic model to obtain at least one word; respectively acquire the characteristics of the at least one word;

S23. Determine the amount of information contained in the at least one word according to the characteristics, and select at least one word containing a large amount of information from the at least one word as a keyword;

S24. Make a window centering on the keyword respectively, and determine the context words of the keyword;

S25. Match the context words and standard synonyms with the colloquial word lexicon to obtain a matching result;

S26. Analyze semantics according to the matching result to obtain first semantic information.

Compared with related technologies, the semantic analysis method based on knowledge map provided by the present invention has the following

Beneficial effect:

1. The present invention recognizes the text information and pitch information in the speech data according to the pre-trained semantic model, and analyzes it through standard synonyms and colloquial word lexicons to obtain the first semantic information, according to The first semantic information is combined with the constructed knowledge map to perform association analysis to obtain the second semantic information. By comparing the second semantic information with the first semantic information, the final output voice information is obtained. By comparing the dual voice information, a more consistent Semantic information required by users to improve the accuracy of information analysis in semantic data;

2. Before performing semantic analysis, the present invention first constructs a knowledge map, analyzes the data source, and generates a plurality of semantic vectors by analyzing it; performs data fusion on the generated semantic vectors, and performs data fusion on the fused semantic vectors. Mining the association relationship between semantic vectors, generating the relationship between semantic vectors, forming a semantic database for the fused semantic vector and the relationship between semantic vectors, and storing it in the server, and evaluating the quality of the semantic database to complete the construction of the knowledge graph , through the relationship between semantic vectors, accurate semantic analysis can be realized;

Description of drawings

Fig. 1 is a schematic flow chart of a semantic analysis method based on a knowledge map provided by the present invention;

FIG. 2 is a schematic flowchart of a specific method for constructing the knowledge map shown in FIG. 1;

Fig. 3 is the flow schematic diagram that obtains text information from speech data shown in Fig. 1;

Fig. 4 is the flow diagram that obtains tone information from voice data shown in Fig. 1;

FIG. 5 is a schematic flow diagram of obtaining first semantic information from text information shown in FIG. 1 .

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments It is a part of embodiments of the present invention, but not all embodiments. The components of the embodiments of the invention generally described and illustrated in the figures herein may be arranged and designed in a variety of different configurations.

Accordingly, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

It should be noted that like numerals and letters denote similar items in the following figures, therefore, once an item is defined in one figure, it does not require further definition and explanation in subsequent figures.

In the description of the embodiments of the present invention, it should be noted that if the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", The orientation or positional relationship indicated by "outside" is based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship that is usually placed when the product of the invention is used, and is only for the convenience of describing the present invention and simplifying the description. It is not to indicate or imply that the device or element referred to must have a particular orientation, be constructed in a particular orientation, or operate in a particular orientation, and thus should not be construed as limiting the invention. In addition, the terms "first", "second", "third" and so on are only used for distinguishing descriptions, and should not be understood as indicating or implying relative importance.

In addition, the appearance of the terms "horizontal", "vertical", "overhanging" etc. does not mean that the parts are absolutely horizontal or overhanging, but may be slightly inclined. For example, "horizontal" only means that its direction is more horizontal than "vertical", and it does not mean that the structure must be completely horizontal, but can be slightly inclined.

In the description of the embodiments of the present invention, it should also be noted that, unless otherwise specified and limited, the terms "setting", "installation", "connection" and "connection" should be interpreted in a broad sense, for example, It can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediary, and it can be the internal communication of two components. Those of ordinary skill in the art can understand the specific meanings of the above terms in the present invention according to specific situations.

The specific implementation of the present invention will be described in detail below in conjunction with specific embodiments.

The semantic analysis method based on the knowledge map provided by the present invention includes: the specific steps are as follows:

S1. Acquire voice data, and screen the voice data, so as to eliminate messy information, leaving only effective sentence patterns, and analyze the effective sentence patterns to obtain text information and tone information, and combine the obtained text information and tone information input into the speech model;

S2. Recognize the text information and tone information according to the pre-trained semantic model, and analyze them through the standard synonyms and colloquial word lexicon to obtain the first semantic information;

S3. Perform association analysis based on the first semantic information combined with the constructed knowledge map to obtain second semantic information;

S4. Comparing the second semantic information with the first semantic information, when the difference value of the comparison result is lower than the set value, determine the semantic analysis result according to the second semantic information, and output it, and compare the difference value of the result When it is higher, the user is reminded to re-enter.

It should be noted that the acquisition of voice data can generally be obtained through sound acquisition equipment, such as microphones such as microphones. The user speaks into the microphone so that the microphone can collect the voice data of the user's speech. The voice data can also be directly collected through audio collection devices, etc. get. For example, directly collect an audio file to obtain speech data.

It is further explained that by analyzing the effective sentence patterns, the text corresponding to each audio is obtained, and these texts are combined, and the combined results are rehearsed in order, and the rehearsed information is generated into text information, which can be used for analysis. Obtain the meaning of what the user said. By inputting the text information and tone information into the semantic model, the combination of tone order and tone information that is too far apart can be eliminated through the tone information, and then combined with the standard synonyms and colloquial word lexicon, eliminate It is impossible to form a combination of words before and after, and obtain the corresponding first semantic information, which can be illustrated by the following examples:

By removing the messy information from the voice data, only the effective sentence pattern ni de quan li yao qiu shishen me is left, and then by obtaining the text corresponding to each audio, and combining the text in order, we can get "your persuasion What is the requirement", "What is your full demand", "What is your claim" and other groups of words, and the tone information is obtained. The tone order of the effective sentence pattern is three tones, soft tones, two tones, and four tones. sound, one tone, two tone, four tone, two tone, soft tone; i.e. nǐde quán lìyāo qiúshìshén me, combined with the standard synonyms and colloquial word lexicon, the first semantic information "what is your claim" can be obtained, and Combined with the analysis of the knowledge map, since the knowledge map contains the user's behavior data, the second semantic information "what is your claim (book)" is obtained, that is, the question to be asked is what is the patent claim, not an inquiry Claims in the legal sense, and by comparing the second semantic information "what is your claim (book)" with the first semantic information "what is your claim", the difference value is obtained: "(book)" , and "(book)" is lower than the set value, and the set value can be the number of words, so the second second semantic information "what is your claim (book)" is output.

In an embodiment of the present invention, the specific method of constructing the knowledge map is as follows:

S31. Connect various data sources in a visual way, verify and manage the data sources, and establish a basic knowledge base according to the data sources;

S32. Analyze the data source, and generate multiple semantic vectors by analyzing it;

S33. Perform data fusion on the generated semantic vectors, and mine association relationships between the fused semantic vectors to generate relationships between semantic vectors;

S34, forming a semantic database from the fused semantic vector and the relationship between the semantic vectors, storing it in the server, evaluating the quality of the semantic database, and completing the construction of the knowledge map.

In an embodiment of the present invention, the data sources in step S31 include standard synonyms and colloquial word thesaurus, basic information filled in by the user, user's usual behavior data, user's chat information and web page data published on the Internet.

It should be noted that standard synonyms and colloquial word thesaurus do not need to be processed, and are directly added to the knowledge graph system. Except for a few fields, the basic information filled in by users needs to be further processed, and many fields can be directly used for modeling or added to In the knowledge graph system, for behavioral data and user chat information, some simple processing is required to extract effective information such as "the length of time the user stays on a certain page", "the user's usual words", etc. For the network For webpage data published on the Internet, data processing needs to be carried out through information extraction related technologies.

In an embodiment of the present invention, the speech data is obtained, and the speech data is screened to remove the messy information, leaving only valid sentence patterns, and analyzing the effective sentence patterns to obtain the text information. The steps are as follows:

S11. Extracting sound features from the speech data;

S12. Match the sound feature quantity with the modeled sound data in the sound library, so as to obtain sound data with similarity matching;

S13. Compare and match the voice data with the voice data stored in the text and voice database to obtain text information.

In an embodiment of the present invention, the text and speech database stores text, speech corresponding to the text, and speech data of extended words and sentences of the text.

In an embodiment of the present invention, the voice data is obtained, and the voice data is screened to remove messy information, leaving only valid sentence patterns, and analyzing the effective sentence patterns to obtain the tone information. The steps are as follows:

S14. Perform spectrum analysis on the voice data, and extract tone phonemes from the voice data;

S15. Match the tone of the voice data in the tone model according to the tone phoneme.

In the embodiment of the present invention, the text information and tone information are identified according to the pre-trained semantic model, and analyzed through the standard synonyms and colloquial word lexicon to obtain the first semantic information. :

S21. Identify the tone information through the voice model, and obtain the tone order, and eliminate the combination of the tone order in the text information and the recognized tone order tone information through the tone information;

Since not everyone in the voice data is pronounced as a standard syllable, a difference value can be set, and when the difference value is too large, the data will be eliminated.

S22. Segment the text information through a semantic model to obtain at least one word; respectively acquire the characteristics of the at least one word;

The characteristics of words can be divided into nouns, verbs _, adjectives, numerals, quantifiers, pronouns, adverbs, prepositions, conjunctions, auxiliary words, interjections and onomatopoeia, and continue to be subdivided into several small categories within the category. For example, nouns can be divided into person nouns, thing nouns, time nouns, location nouns, relative nouns, etc. By acquiring the characteristics of the words, the meaning of the words can be better understood; the details are as follows: your (personal pronoun) ( Structural particle) right (thing noun) requires (thing noun) what (interrogative pronoun).

S23. Determine the amount of information contained in the at least one word according to the characteristics, and select at least one word containing a large amount of information from the at least one word as a keyword;

S24. Make a window centering on the keyword respectively, and determine the context words of the keyword;

S25. Match the context words and standard synonyms with the colloquial word lexicon to obtain a matching result;

S26. Analyze semantics according to the matching result to obtain first semantic information.

The circuits and controls involved in the present invention are all prior art, and will not be repeated here.

The above is only an embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the description of the present invention and the contents of the accompanying drawings, or directly or indirectly used in other related technologies fields, are all included in the scope of patent protection of the present invention in the same way.

Claims (7)

1. A semantic analysis method based on a knowledge map, characterized in that the specific steps are as follows: S1. Acquire voice data, and screen the voice data, so as to eliminate messy information, leaving only effective sentence patterns, and analyze the effective sentence patterns to obtain text information and tone information, and combine the obtained text information and tone information input into the speech model; S2. Recognize the text information and tone information according to the pre-trained semantic model, and analyze them through the standard synonyms and colloquial word lexicon to obtain the first semantic information; S3. Perform association analysis based on the first semantic information combined with the constructed knowledge map to obtain second semantic information; S4. Comparing the second semantic information with the first semantic information, when the difference value of the comparison result is lower than the set value, determine the semantic analysis result according to the second semantic information, and output it, and compare the difference value of the result When it is higher, the user is reminded to re-enter. 2. The semantic analysis method based on knowledge map according to claim 1, characterized in that, the specific method of the construction of said knowledge map is: S31. Connect various data sources in a visual way, verify and manage the data sources, and establish a basic knowledge base according to the data sources; S32. Analyze the data source, and generate multiple semantic vectors by analyzing it; S33. Perform data fusion on the generated semantic vectors, and mine association relationships between the fused semantic vectors to generate relationships between semantic vectors; S34, forming a semantic database from the fused semantic vector and the relationship between the semantic vectors, storing it in the server, evaluating the quality of the semantic database, and completing the construction of the knowledge map. 3. The semantic analysis method based on knowledge graph according to claim 2, characterized in that, the data sources in the step S31 include standard synonyms and colloquial word lexicons, basic information filled in by users, and user's usual behavior data , user's chat information and webpage data published on the Internet. 4. The semantic analysis method based on knowledge graph according to claim 1, characterized in that, the speech data is obtained, and the speech data is screened, so as to eliminate the messy information, leaving only valid sentence patterns, and analyzing the effective sentence patterns , the steps to get the text information are: S11. Extracting sound features from the speech data; S12. Match the sound feature quantity with the modeled sound data in the sound library, so as to obtain sound data with similarity matching; S13. Compare and match the voice data with the voice data stored in the text and voice database to obtain text information. 5. The semantic analysis method based on knowledge graph according to claim 4, characterized in that, the text and speech database stores text, speech corresponding to the text and speech data of extended words and sentences of the text. 6. The semantic analysis method based on knowledge graph according to claim 1, characterized in that, the speech data is obtained, and the speech data is screened, so as to eliminate the messy information, leaving only valid sentence patterns, and analyzing the effective sentence patterns , the steps to obtain pitch information are: S14. Perform spectrum analysis on the voice data, and extract tone phonemes from the voice data; S15. Match the tone of the voice data in the tone model according to the tone phoneme. 7. The semantic analysis method based on knowledge graph according to claim 1, characterized in that, according to the pre-trained semantic model, text information and tone information are identified, and through standard synonyms and colloquial word lexicon, the The steps of analyzing it to obtain the first semantic information are: S21. Identify the tone information through the voice model, and obtain the tone order, and eliminate the combination of the tone order in the text information and the recognized tone order tone information through the tone information; S22. Segment the text information through a semantic model to obtain at least one word; respectively acquire the characteristics of the at least one word; S23. Determine the amount of information contained in the at least one word according to the characteristics, and select at least one word containing a large amount of information from the at least one word as a keyword; S24. Make a window centering on the keyword respectively, and determine the context words of the keyword; S25. Match the context words and standard synonyms with the colloquial word lexicon to obtain a matching result; S26. Analyze semantics according to the matching result to obtain first semantic information.