WO2023035529A1

WO2023035529A1 - Intent recognition-based information intelligent query method and apparatus, device and medium

Info

Publication number: WO2023035529A1
Application number: PCT/CN2022/071789
Authority: WO
Inventors: 舒畅; 陈又新
Original assignee: 平安科技（深圳）有限公司
Priority date: 2021-09-10
Filing date: 2022-01-13
Publication date: 2023-03-16
Also published as: CN113761206A

Abstract

The present application discloses an information intelligent query method and apparatus based on intent recognition, a device and a medium. The method comprises: extracting and obtaining corresponding target text information from query request information inputted by a user; parsing the target text information according to a text intent parsing model to obtain an intent type; according to the intent type and a relationship recognition network, recognizing the association between feature words in the target text information to obtain feature word association information; and querying an information database according to the intent type and the feature word association information to obtain an information query result.

Description

Information intelligent query method, device, equipment and medium based on intent recognition

This application claims the priority of the Chinese patent application with the application number 202111060715.2 submitted to the China Patent Office on September 10, 2021, and the title of the invention is "Intent Recognition-Based Information Intelligent Query Method, Device, Equipment, and Medium", the entire content of which Incorporated in this application by reference.

technical field

This application relates to the field of data query technology, and belongs to the application scenario of intelligent information query based on intention recognition in smart education, and in particular to an information intelligent query method, device, equipment and medium based on intention recognition.

Background technique

With the rapid development of Internet technology, the intelligence of information processing is also getting higher and higher. When enterprises reply to customers' information, they usually perform semantic analysis based on the user's questions, and determine the corresponding keywords based on the semantic analysis, and query the data information contained in the database based on the keywords to obtain the corresponding query results. However, the inventors found that the same keyword can obtain multiple different query results under multiple intentions. Accurate query results corresponding to the query intent are obtained from the results for feedback, which greatly affects the accuracy of information query.

Contents of the invention

Embodiments of the present application provide an intelligent information query method, device, device, and medium based on intent recognition, aiming to solve the problem of low accuracy of information query in existing methods.

In the first aspect, the embodiment of the present application provides an intelligent information query method based on intent recognition, which includes:

If the input query request information is received, the corresponding target text information is extracted from the query request information;

Analyzing the target text information according to a preset text intent analysis model to obtain a corresponding intent type;

According to the preset relationship recognition network and the intent type, the association relationship between the feature words in the target text information is identified, and the feature word association information corresponding to the target text information is obtained;

A pre-stored information database is queried according to the intent type and the associated information of the characteristic words, so as to obtain an information query result corresponding to the query request information.

In the second aspect, the embodiment of the present application provides an intelligent information query device based on intention recognition, which includes:

A target text information acquisition unit, configured to extract corresponding target text information from the query request information if the input query request information is received;

An intent type identification unit, configured to analyze the target text information according to a preset text intent analysis model to obtain a corresponding intent type;

A feature word association information acquisition unit, configured to identify the association relationship between the feature words in the target text information according to the preset relationship recognition network and the intent type, and obtain the feature word association information corresponding to the target text information ;

An information query result acquisition unit, configured to query a prestored information database according to the intent type and the feature word association information, so as to obtain an information query result corresponding to the query request information.

In the third aspect, the embodiment of the present application further provides a computer device, which includes a memory, a processor, and a computer program stored on the memory and operable on the processor, and the processor executes the computer program. The program implements the intelligent information query method based on intent recognition described in the first aspect above.

In a fourth aspect, the embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the above-mentioned first step. In one aspect, the intelligent query method for information based on intent recognition.

Embodiments of the present application provide a method, device, device, and medium for intelligent information query based on intent recognition. Extract the corresponding target text information from the query request information input by the user, analyze the target text information according to the text intent analysis model to obtain the intent type, and identify the association between feature words in the target text information according to the intent type and relationship recognition network According to the relational information of the characteristic words, the information database is queried according to the type of intent and the association information of the characteristic words to obtain the information query results. Through the above method, the accurate intention type can be obtained by analyzing the target text information, and the association relationship between the characteristic words in the target text information can be identified, and the information query result can be obtained based on the association relationship between the intention type and the characteristic words, which can greatly The accuracy of information inquiry is increased, and the efficiency of information inquiry is improved at the same time.

Description of drawings

In order to illustrate the technical solutions of the embodiments of the present application more clearly, the drawings that need to be used in the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present application. Ordinary technicians can also obtain other drawings based on these drawings on the premise of not paying creative work.

FIG. 1 is a schematic flow diagram of an information intelligent query method based on intent recognition provided by an embodiment of the present application;

Fig. 2 is a schematic sub-flow diagram of an information intelligent query method based on intent recognition provided by an embodiment of the present application;

FIG. 3 is a schematic diagram of another sub-flow of the intent recognition-based intelligent information query method provided by the embodiment of the present application;

FIG. 4 is a schematic diagram of another sub-flow of the intelligent information query method based on intent recognition provided by the embodiment of the present application;

FIG. 5 is a schematic diagram of another sub-flow of the intelligent information query method based on intent recognition provided by the embodiment of the present application;

FIG. 6 is a schematic diagram of another sub-flow of the intelligent information query method based on intent recognition provided by the embodiment of the present application;

FIG. 7 is another schematic flowchart of an intelligent information query method based on intent recognition provided by an embodiment of the present application;

FIG. 8 is a schematic block diagram of an information intelligent query device based on intent recognition provided by an embodiment of the present application;

Fig. 9 is a schematic block diagram of a computer device provided by an embodiment of the present application.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of them. Based on the embodiments in this application, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the scope of protection of this application.

It should be understood that when used in this specification and the appended claims, the terms "comprising" and "comprises" indicate the presence of described features, integers, steps, operations, elements and/or components, but do not exclude one or Presence or addition of multiple other features, integers, steps, operations, elements, components and/or collections thereof.

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

It should also be further understood that the term "and/or" used in the description of the present application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations .

Please refer to Fig. 1, Fig. 1 is a schematic flow chart of an intelligent information query method based on intent recognition provided by an embodiment of the present application; the intelligent query method for information based on intent recognition is applied to a user terminal or a management server, the The intelligent query method is executed through the application software installed in the user terminal or the management server. The user terminal is a terminal device that can receive the query request information input by the user and perform information query to obtain the query results, such as desktop computers, notebook computers, and tablets. For computers or mobile phones, etc., the management server is a server that can be used to receive query request information from user terminals, perform information query, obtain information query results, and provide feedback, such as servers built by enterprises or government departments. As shown in FIG. 1, the method includes steps S110-S140.

S110. If the input query request information is received, extract corresponding target text information from the query request information.

The query request information can be sent to the management server by the user of the user terminal through the user terminal, or can be input by the user into the user terminal. The query request information can be text, voice or short video, and the corresponding target needs to be obtained from the query request information. Text information, and obtain the user's real information query intention based on the target text information.

In one embodiment, as shown in FIG. 2 , step S110 includes sub-steps S111 , S112 and S113 .

S111. Determine the information type of the query request information.

Wherein, the information type includes a text information type and a voice information type. Specifically, the query request information includes corresponding format identification information. The format identification information is information used to identify the format of the query request information. The information type of the query request information can be determined by the format identification information of the query request information.

For example, if the format identification information is txt, xml or string, the corresponding query request information is text information type; if the format identification information is wav, mp3, wma, avi, flv, rmvb, then the corresponding query request information is voice Information type (including audio information type and video information type).

For example, the user of the user terminal enters text in the question box on the terminal page and clicks the confirmation button, then the user terminal receives the text information to obtain query request information, or sends the text to the management server as query request information; The voice input button on the page, say your own question and click the confirmation button, then the user terminal receives the voice to get the query request information, or sends the recorded voice as the query request information to the management server; the user clicks the video input button on the terminal page button, speak your question to the video capture device of the user terminal and click the confirm button, then the user terminal receives the recorded segment of video to get the query request information, or sends the recorded short video as the query request information to the management server .

S112. If the information type of the query request information is a voice information type, recognize the voice information included in the query request information according to a pre-stored voice recognition model to obtain target text information corresponding to the query request information.

If the query request information is of voice information type, the query request information may be of audio information type or video information type, both of which include voice information. The speech recognition model is a model that recognizes and converts speech information included in audio information or video information, wherein the speech recognition model includes a noise judgment rule and a text information acquisition model. The noise judgment rule is the rule for judging whether the speech information contains noise. The text information acquisition model is the model for obtaining the corresponding text information from the speech information. If the speech information contains noise, it will affect the information obtained from the speech information. Therefore, before obtaining the corresponding target text information from the voice information, the noise judgment rule can be used to judge whether the voice information contains noise, so as to ensure that more accurate information can be obtained from the noise-free voice information. Target text information.

In one embodiment, as shown in FIG. 3 , step S112 includes sub-steps S1121 , S1122 and S1123 .

S1121. Determine whether the voice information in the query request information contains noise according to the noise determination rule.

It is judged according to the noise judgment rule whether the voice information in the query request information contains noise. Specifically, since the frequency of human speech is in a fixed frequency range (85Hz-1100Hz), the voice information can be sampled based on the frequency of the voiceprint signal in the voice information. The sampling frequency can be 5-50Hz, and the sampling frequency is The number of times the loudness value of the voiceprint signal is obtained from the voice information per second, and the unit of the loudness value is decibel (dB). For example, if the sampling frequency is 10 Hz, the loudness value in the frequency range corresponding to the voiceprint signal is obtained from the voice information 10 times per second. Based on the above-mentioned sampling method, the average value of multiple loudness values in the above-mentioned fixed frequency interval can be obtained from the speech information as the target sound signal strength, and the above-mentioned sampling method can be used to obtain the multiplicity of other sound signals not in the above-mentioned fixed frequency interval from the speech information. The average value of the loudness values is used as the background noise signal strength, and it is judged whether the ratio between the background noise signal strength and the target sound signal strength is greater than the ratio threshold preset in the noise judgment rule, and if the ratio is greater than the ratio threshold value, the query request information is judged The voice information in contains noise; if the ratio is not greater than the ratio threshold, it is judged that the voice information in the query request information does not contain noise.

S1122. If the voice information in the query request information does not contain noise, identify the voice information in the query request information according to the text information acquisition model, so as to obtain target text information corresponding to the query request information.

If the voice information in the query request information does not contain noise, identify the voice information in the query request information according to the text information acquisition model, so as to obtain target text information corresponding to the query request information. If the speech information in the query request information does not contain noise, the speech information can be recognized according to the text information acquisition model to obtain the corresponding target text information. Specifically, the text information acquisition model includes an acoustic model, a speech feature dictionary and a semantic analysis dictionary .

In one embodiment, step S1122 includes a sub-step: segmenting the voice information in the query request information according to the acoustic model in the text information acquisition model to obtain a plurality of phonemes contained in the voice information; Match the phonemes according to the phonetic feature dictionary in the text information acquisition model to convert the phonemes into pinyin information; perform semantic analysis on the pinyin information according to the semantic analysis dictionary in the text information acquisition model to obtain Target text information corresponding to the query request information.

The voice information in the query request information is segmented according to the acoustic model in the text information acquisition model to obtain a plurality of phonemes contained in the voice information. Specifically, the voice information included in the audio information or video information is composed of phonemes that are pronounced by multiple characters, and the phonemes of a character include the frequency and timbre of the character's pronunciation. The acoustic model contains the phonemes of all character pronunciations. By matching the voice information with all the phonemes in the acoustic model, the phonemes of a single character in the voice information can be segmented, and finally the voice information of the query request information can be obtained through segmentation. Multiple phonemes included.

The phonemes are matched according to the phonetic feature dictionary in the text information acquisition model to convert the phonemes into pinyin information. The phonetic feature dictionary contains the phoneme information corresponding to all characters’ pinyin. By matching the obtained phoneme with the phoneme information corresponding to the character’s pinyin, the phoneme of a single character can be converted into the character’s pinyin matching the phoneme in the phonetic feature dictionary , so as to convert all the phonemes contained in the speech information into pinyin information.

Perform semantic analysis on the pinyin information according to the semantic analysis dictionary in the text information acquisition model to obtain target text information corresponding to the query request information. The semantic analysis dictionary contains the mapping relationship between a single pinyin information or multiple pinyin phrases and text information. Through the mapping relationship contained in the semantic analysis dictionary, the obtained pinyin information can be semantically analyzed to convert the pinyin information into corresponding target text message for .

S1123. If the voice information in the query request information contains noise, feed back re-input prompt information to prompt re-input of the query request information.

If the voice information in the query request information contains noise, it will affect the accuracy of the acquired target text information. At this time, the re-input prompt information can be fed back to prompt the user to re-enter the query request information in a low-noise environment.

S113. If the information type of the query request information is text information, determine the query request information as target text information.

If the query request information is of text information type, the query request information is used as the target text information. If the query request information is of text information type, there is no need to process the query request information, and the query request information can be directly used as target text information for subsequent processing.

S120. Analyze the target text information according to a preset text intent analysis model to obtain a corresponding intent type.

The target text information is parsed according to a preset text intent parsing model to obtain a corresponding intent type. The text intent parsing model includes word segmentation processing rules and a classification neural network. The target text information can be analyzed through the text intent analysis model to obtain the intent type of the target text information.

In a more specific embodiment, as shown in FIG. 4, step S120 includes sub-steps S121 and S122.

S121. Perform word segmentation processing on the target text information according to the word segmentation processing rule to obtain a word segmentation result.

Perform word segmentation processing on the target text information according to the word segmentation processing rule to obtain a word segmentation result. The word segmentation processing can be performed on the target text information according to the word segmentation processing rules to obtain the corresponding word segmentation results, which include feature words composed of one or more characters and attribute information of each feature word. The word segmentation processing rules include a vector conversion database and a part-of-speech tagging network.

In a more specific embodiment, as shown in FIG. 5, step S121 includes sub-steps S1211, S1212 and S1213.

S1211. Obtain a corresponding character feature vector from the target text information according to the vector conversion database.

The vector conversion database is a data table that stores the relationship between characters and unit feature vectors. The vector conversion database contains a unit feature vector corresponding to each character, and the unit feature vector can be used to quantify the characteristics of characters. According to the target text information, a unit feature vector corresponding to each character in the target text information can be obtained from the vector conversion database, and the unit feature vectors corresponding to multiple characters contained in the target text information can be combined to obtain A corresponding text feature vector, that is, converting the target text information into a corresponding text feature vector.

For example, if the target text information is "Is there an English class tomorrow afternoon?", the corresponding text feature vectors are [101, 9251, 3151, 7821, 6215, 4152, 3324, 3620, 8512, 4831, 8514, 0635, 102] . Among them, "101" is the unit feature vector at the beginning of the sentence, and "102" is the unit feature vector at the end of the sentence.

S1212. Perform part-of-speech tagging on the text feature vector according to the part-of-speech tagging network to obtain part-of-speech tagging information corresponding to each character in the target text information.

The part-of-speech tagging information corresponding to each character in the target text information can be obtained by performing part-of-speech tagging on the text feature vector through the part-of-speech tagging network. Specifically, the part-of-speech tagging network can be obtained based on the combination of BERT (Bidirectional Encoder Representations from Transformers) neural network and CRF (Intro2 Conditional Random Field) neural network. The BERT neural network can perform feature calculation on the text feature vector to obtain the feature array corresponding to the text feature vector. If the text feature vector is a 1×H-dimensional vector, the obtained feature array contains L×H (L rows and H columns) features An array of values, each feature value belongs to the value range [0, 1]. The CRF neural network is a serialized labeling algorithm network. Input the feature array into the CRF neural network to get the label value corresponding to the feature sequence of each row in the feature array, except for the label value corresponding to the unit feature vector at the beginning of the sentence and the unit feature vector at the end of the sentence. , the remaining multiple tag values correspond to the characters contained in the target text information, that is, the part-of-speech tag information corresponding to the target text information is obtained.

S1213. Perform word segmentation processing on the target text information according to the part-of-speech tagging information corresponding to each character, to obtain a corresponding word segmentation result.

According to the character label information, word segmentation can be performed on the characters contained in the target text information, and it is judged whether the label values of two adjacent characters are the same. If they are the same, the adjacent characters are combined; characters for word segmentation. Through the above method, word segmentation processing can be performed on all characters contained in the target text information, and a word segmentation result can be obtained.

For example, the label value of "Ming" is "203: noun-time", and the label value of "day" is "203: noun-time", and the two characters contain "noun-time" at the same time, then the two characters can be combined , and use the common label value as the label value of the phrase obtained after combination.

S122. Input the word segmentation result and the text feature vector into the classification neural network, so as to obtain the intent type corresponding to the target text information.

Combine the label value of each character or phrase in the word segmentation result with the text feature vector to obtain the combined feature, and input it into the classification neural network to obtain the corresponding intent type. The classification neural network can be composed of an input layer, multiple intermediate layers and an output layer, between the input layer and the first intermediate layer, between the intermediate layer and other adjacent intermediate layers, between the last intermediate layer and the output layer The association is performed through an association formula. For example, a certain association formula can be expressed as y=r×x+t, and r and t are the parameter values in the association formula. The number of input nodes contained in the input layer is not less than the feature value contained in the combined feature, and the output layer contains multiple output nodes, each output node corresponds to a classification type, and the combined feature is input into the classification neural network for calculation. Obtain the corresponding output result from the output layer, the output result contains the matching probability corresponding to each output node, and the classification type corresponding to an output node with the highest matching probability can be obtained as the intent type corresponding to the target text information.

For example, the label value of "Ming" is "203", and its corresponding unit feature vector in the text feature vector is "9251", then the label value and the unit feature vector are combined as the combined feature corresponding to "Ming". , obtain the combination feature corresponding to each character in the target text information and input it into the classification neural network to obtain the corresponding intent classification.

S130. Perform association relationship identification between feature words in the target text information according to the preset relationship identification network and the intent type, and obtain feature word association information corresponding to the target text information.

Word segmentation is performed on the target text information to obtain a word segmentation result, which contains a single character or phrase obtained from the word segmentation, and the single character or phrase is used as a feature word contained in the target text information, and based on the relationship recognition network and the obtained intent type. The feature words are identified for the association relationship, and the feature word association information of whether there is an association relationship between the feature words is obtained.

Specifically, the semantic matrix is constructed based on the feature words contained in the target text information, and the association relationship between the feature words is identified through a fully connected network. First, the intent type is converted into an intent feature vector, and each of the feature words is The unit feature vector of the character in the text feature vector is combined with the label value of the character to obtain the feature word vector of each feature word, and the feature word vector and intention feature vector corresponding to the two feature words are input into the fully connected network at the same time, and based on The normalization function Softmax(f(a,b)) normalizes the output results of the fully connected network to obtain the normalized value, and judges whether the normalized value is greater than the preset threshold, then the two feature words Whether there is an association relationship between them is identified. If the normalized value is greater than the preset threshold, there is an association between the two feature words; if the normalized value is not greater than the preset threshold, there is no relationship between the two feature words. connection relation.

For example, the semantic matrix constructed based on the feature words contained in "Is there an English class tomorrow afternoon?" is shown in Table 1.

Table 1

If the characteristic word "Tomorrow afternoon/N-Time" has an association relationship with the characteristic word "English class/OBJ-Course", the corresponding association relationship can be identified in the semantic matrix.

S140. Query a pre-stored information database according to the intent type and the associated information of the characteristic words, so as to obtain an information query result corresponding to the query request information.

Data query can be performed in the pre-stored information database according to the type of intent and the association relationship of feature words, so as to obtain the information query result corresponding to the query request from the information database. Wherein, the information database is a structured database for recording various information pre-stored in the management server or the user terminal. For example, in the scenario of online education, the information database may contain information related to course schedules and course fees, as well as information related to course exercises, articles, or knowledge points.

In a more specific embodiment, as shown in FIG. 6, step S140 includes sub-steps S141, S142 and S143.

S141. Generate a corresponding associated query sentence according to the plurality of associated characteristic words in the characteristic word association information and the intent type.

If there is an association relationship between two or more feature words in the feature word association information, a corresponding associated query statement can be generated based on the associated multiple feature words and intent types, and the associated query statement can be an SQL query statement.

For example, if the intent type is "scheduling courses", and there is an association relationship between the feature words "tomorrow afternoon/N-Time" and "English class/OBJ-Course", then the corresponding associated query statement can be generated based on the above information, and the associated query statement It can be used to reflect: Inquire "English Class/OBJ-Course" of "Tomorrow afternoon/N-Time" in "Class Arrangement Information (Course Schedule Information)".

S142. Generate a separate query statement according to the feature words that do not have an association relationship in the feature word association relationship and the intent type.

If the associated feature information contains feature words that do not have an association relationship, a separate query statement can be generated based on the feature words and the intent type, and the generated separate query statement can also be an SQL query statement.

For example, if the intent type is "purchase-pay" and the feature word is "writing class/OBJ-Course", then a corresponding separate query statement can be generated based on the above information, and the separate query statement can be used to reflect: In "Payment information (course fee )" to query "Writing Course/OBJ-Course".

The query statement generated based on the intent type and the associated information of the characteristic words may only include the associated query statement, or only the individual query statement, or both the associated query statement and the individual query statement.

S143. Perform a query on the information database according to the associated query statement and/or the individual query statement, so as to obtain a corresponding information query result.

The information database can be queried according to the obtained query statements. Each query statement can realize a query of the information database, and each query statement can obtain a query result. Based on the generated one or more query statements, the information The database is queried, and the obtained query results are synthesized to obtain the information query results corresponding to the query request information

In one embodiment, as shown in FIG. 7 , step S150 is further included after step S140 .

S150. Upload the information query result to the blockchain.

Upload the information query result to the block chain. The corresponding abstract information is obtained based on the information query result. Specifically, the abstract information is obtained by hashing the information query result, for example, by using the sha256s algorithm. Uploading summary information to the blockchain guarantees its security and fairness and transparency to users. The user device can download the summary information from the blockchain to verify whether the information query result has been tampered with. The blockchain referred to in this example is a new application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm. Blockchain (Blockchain), essentially a decentralized database, is a series of data blocks associated with each other using cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify its Validity of information (anti-counterfeiting) and generation of the next block. The blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.

This application can be applied to smart education and other scenarios where information is intelligently searched based on intent recognition, thereby promoting the construction of smart education.

In the intelligent information query method based on intent recognition provided in the embodiment of the present application, the corresponding target text information is extracted from the query request information input by the user, and the target text information is analyzed according to the text intent analysis model to obtain the intent type, and According to the type of intent and the relationship recognition network, the association relationship between the feature words in the target text information is identified to obtain the associated information of the feature words, and the information database is queried according to the type of intent and the associated information of the feature words to obtain the information query result. Through the above method, the accurate intention type can be obtained by analyzing the target text information, and the association relationship between the characteristic words in the target text information can be identified, and the information query result can be obtained based on the association relationship between the intention type and the characteristic words, which can greatly The accuracy of information inquiry is increased, and the efficiency of information inquiry is improved at the same time.

The embodiment of the present application also provides an intelligent information query device based on intention recognition, which is used to implement any embodiment of the aforementioned intelligent information query method based on intention recognition. Specifically, please refer to Fig. 8, Fig. 8 is a schematic block diagram of an information intelligent query device based on intention recognition provided by an embodiment of the present application.

As shown in FIG. 8 , the intelligent information query device 100 based on intention recognition includes a target text information acquisition unit 110 , an intention type identification unit 120 , a characteristic word related information acquisition unit 130 and an information query result acquisition unit 140 .

The target text information acquisition unit 110 is configured to extract corresponding target text information from the query request information if the input query request information is received.

In an embodiment, the target text information acquisition unit 110 includes subunits: a type judgment unit, a speech information recognition unit, and a target text information determination unit.

A type judging unit, configured to judge the information type of the query request information.

The voice information recognition unit is configured to, if the query request information is not text information, recognize the voice information contained in the query request information according to a pre-stored speech recognition model to obtain target text information corresponding to the query request information.

In one embodiment, the index data set acquisition unit includes a subunit: a noise judgment unit, configured to judge whether the speech information in the query request information contains noise according to the noise judgment rule; a speech recognition unit, using If the voice information in the query request information does not contain noise, identify the voice information in the query request information according to the text information acquisition model to obtain the target text information corresponding to the query request information; prompt An information feedback unit, configured to feed back re-input prompt information to prompt re-input of the query request information if the voice information in the query request information contains noise.

The target text information determining unit is configured to determine the query request information as the target text information if the information type of the query request information is a text information type.

The intent type identification unit 120 is configured to analyze the target text information according to a preset text intent analysis model to obtain the corresponding intent type.

In one embodiment, the intent type identification unit 120 includes subunits: a word segmentation processing unit, configured to perform word segmentation processing on the target text information according to the word segmentation processing rules to obtain a word segmentation result; a classification unit, configured to classify the The word segmentation result and the character feature vector are input into the classification neural network to obtain the intent type corresponding to the target text information.

In one embodiment, the word segmentation processing unit includes subunits: a character feature vector acquisition unit, configured to acquire a corresponding character feature vector from the target text information according to the vector conversion database; a part-of-speech tagging information acquisition unit, used Perform part-of-speech tagging on the text feature vector according to the part-of-speech tagging network to obtain the part-of-speech tagging information corresponding to each character in the target text information; the word segmentation result acquisition unit is used to pair The target text information is subjected to word segmentation processing to obtain corresponding word segmentation results.

The feature word association information acquisition unit 130 is used to identify the association relationship between the feature words in the target text information according to the preset relationship recognition network and the intent type, and obtain the feature word association corresponding to the target text information information.

The information query result acquisition unit 140 is configured to query a pre-stored information database according to the intent type and the feature word association information, so as to obtain an information query result corresponding to the query request information.

In one embodiment, the information query result acquisition unit 140 includes a subunit: an associated query statement generation unit, configured to generate a corresponding query statement according to a plurality of associated characteristic words in the characteristic word association information and the intent type. Associated query statement; a separate query statement generation unit, used to generate a separate query statement according to the feature words and the intent type that do not have an associated relationship in the feature word association relationship; an information query unit, used to generate a separate query statement based on the associated query statement and /or the separate query statement is used to query the information database to obtain a corresponding information query result.

In an embodiment, the intelligent information query device 100 based on intention identification further includes a subunit: an information query result storage unit, configured to upload the information query result to the block chain.

The intelligent information query device based on intent recognition provided in the embodiment of the present application applies the above-mentioned intelligent query method based on intent recognition, extracts the corresponding target text information from the query request information input by the user, and analyzes the target text information according to the text intent analysis model. The text information is analyzed to obtain the intent type, and the relationship between the feature words in the target text information is identified according to the intent type and the relationship recognition network to obtain the feature word association information, and the information database is queried according to the intent type and feature word association information to obtain the information query result. Through the above method, the accurate intention type can be obtained by analyzing the target text information, and the association relationship between the characteristic words in the target text information can be identified, and the information query result can be obtained based on the association relationship between the intention type and the characteristic words, which can greatly The accuracy of information inquiry is increased, and the efficiency of information inquiry is improved at the same time.

The above-mentioned intelligent information query device based on intention recognition can be implemented in the form of a computer program, and the computer program can be run on the computer device as shown in FIG. 9 .

Please refer to FIG. 9 , which is a schematic block diagram of a computer device provided by an embodiment of the present application. The computer device may be a user terminal or a management server for executing the intent recognition-based intelligent information query method to perform intelligent information query based on intent recognition.

Referring to FIG. 9 , the computer device 500 includes a processor 502 connected through a system bus 501 , a memory and a network interface 505 , wherein the memory may include a storage medium 503 and an internal memory 504 .

The storage medium 503 can store an operating system 5031 and a computer program 5032 . When the computer program 5032 is executed, it can cause the processor 502 to execute an intelligent information query method based on intent recognition, wherein the storage medium 503 can be a volatile storage medium or a non-volatile storage medium.

The processor 502 is used to provide calculation and control capabilities and support the operation of the entire computer device 500 .

The internal memory 504 provides an environment for running the computer program 5032 in the storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute an intelligent information query method based on intent recognition.

The network interface 505 is used for network communication, such as providing data transmission and the like. Those skilled in the art can understand that the structure shown in FIG. 9 is only a block diagram of a partial structure related to the solution of this application, and does not constitute a limitation on the computer device 500 on which the solution of this application is applied. The specific computer device 500 may include more or fewer components than shown, or combine certain components, or have a different arrangement of components.

Wherein, the processor 502 is configured to run the computer program 5032 stored in the memory, so as to realize the corresponding functions in the above-mentioned intelligent information query method based on intent recognition.

Those skilled in the art can understand that the embodiment of the computer device shown in FIG. 9 does not constitute a limitation on the specific composition of the computer device. In other embodiments, the computer device may include more or less components than those shown in the illustration. Or combine certain components, or different component arrangements. For example, in some embodiments, the computer device may only include a memory and a processor. In such an embodiment, the structures and functions of the memory and the processor are consistent with those of the embodiment shown in FIG. 9 , and will not be repeated here.

It should be understood that in the embodiment of the present application, the processor 502 may be a central processing unit (Central Processing Unit, CPU), and the processor 502 may also be other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), Application Specific Integrated Circuit (ASIC), off-the-shelf programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Wherein, the general-purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

In another embodiment of the present application a computer readable storage medium is provided. The computer-readable storage medium may be a volatile or non-volatile computer-readable storage medium. The computer-readable storage medium stores a computer program, wherein when the computer program is executed by the processor, the above-mentioned method for intelligent information query based on intention recognition is implemented.

Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described equipment, devices and units can refer to the corresponding process in the foregoing method embodiments, and details are not repeated here. Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the relationship between hardware and software Interchangeability. In the above description, the composition and steps of each example have been generally described according to their functions. Whether these functions are implemented by hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art may use different methods to implement the described functions for each specific application, but such implementation should not be regarded as exceeding the scope of the present application.

In the several embodiments provided in this application, it should be understood that the disclosed devices, devices and methods can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the units is only logical function division. In actual implementation, there may be other division methods, and units with the same function may also be combined into one Units such as multiple units or components may be combined or integrated into another system, or some features may be omitted, or not implemented. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices or units, and may also be electrical, mechanical or other forms of connection.

The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment of the present application.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, each unit may exist separately physically, or two or more units may be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

If the integrated unit is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of software products, and the computer software products are stored in a computer. The read storage medium includes several instructions to make a computer device (which may be a personal computer, a management server, or a network device, etc.) execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned computer-readable storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), magnetic disk or optical disk, and other media that can store program codes.

The above is only a specific embodiment of the application, but the scope of protection of the application is not limited thereto. Any person familiar with the technical field can easily think of various equivalents within the scope of the technology disclosed in the application. Modifications or replacements, these modifications or replacements shall be covered within the scope of protection of this application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

A method for intelligent information query based on intent recognition, comprising:

If the input query request information is received, the corresponding target text information is extracted from the query request information;

Analyzing the target text information according to a preset text intent analysis model to obtain a corresponding intent type;

According to the preset relationship recognition network and the intent type, the association relationship between the feature words in the target text information is identified, and the feature word association information corresponding to the target text information is obtained;

A pre-stored information database is queried according to the intent type and the associated information of the characteristic words, so as to obtain an information query result corresponding to the query request information.
The intelligent information query method based on intent recognition according to claim 1, wherein said extracting corresponding target text information from said query request information comprises:

judging the information type of the query request information;

If the information type of the query request information is a voice information type, then recognize the voice information contained in the query request information according to a pre-stored voice recognition model to obtain target text information corresponding to the query request information;

If the information type of the query request information is a text information type, the query request information is determined as target text information.
The intelligent information query method based on intention recognition according to claim 2, wherein the voice recognition model includes noise judgment rules and text information acquisition models, and the voice information contained in the query request information is analyzed according to the pre-stored voice recognition model. Information is identified to obtain the target text information corresponding to the query request information, including:

judging whether the voice information in the query request information contains noise according to the noise judging rule;

If the voice information in the query request information does not contain noise, identify the voice information in the query request information according to the text information acquisition model, so as to obtain target text information corresponding to the query request information;

If the voice information in the query request information contains noise, feedback re-input prompt information to prompt re-input of the query request information.
The information intelligent query method based on intent recognition according to claim 1, wherein the text intent analysis model includes word segmentation processing rules and a classification neural network, and the target text information is processed according to the preset text intent analysis model Parse to get the corresponding intent type, including:

performing word segmentation processing on the target text information according to the word segmentation processing rules to obtain word segmentation results;

Input the word segmentation result and the character feature vector into the classification neural network to obtain the intent type corresponding to the target text information.
The information intelligent query method based on intent recognition according to claim 4, wherein the word segmentation processing rules include a vector conversion database and a part-of-speech tagging network, and the target text information is subjected to word segmentation processing according to the word segmentation processing rules to obtain Word segmentation results, including:

Obtaining a corresponding character feature vector from the target text information according to the vector conversion database;

Carry out part-of-speech tagging to the character feature vector according to the part-of-speech tagging network, and obtain the part-of-speech tagging information corresponding to each character in the target text information;

The target text information is subjected to word segmentation processing according to the part-of-speech tagging information corresponding to each character, and a corresponding word segmentation result is obtained.
The intelligent information query method based on intent recognition according to claim 1, wherein the pre-stored information database is queried according to the type of intent and the associated information of the feature words to obtain information corresponding to the query request information Information query results, including:

Generate a corresponding associated query statement according to a plurality of associated characteristic words in the characteristic word association information and the intent type;

Generate a separate query statement according to the feature words that do not have an association relationship in the feature word association relationship and the intent type;

According to the associated query statement and/or the individual query statement, the information database is queried to obtain a corresponding information query result.
The intelligent information query method based on intention recognition according to claim 1, further comprising:

Upload the information query result to the block chain.
An information intelligent query device based on intention recognition, comprising:

A target text information acquisition unit, configured to extract corresponding target text information from the query request information if the input query request information is received;

An intent type identification unit, configured to analyze the target text information according to a preset text intent analysis model to obtain a corresponding intent type;

A feature word association information acquisition unit, configured to identify the association relationship between the feature words in the target text information according to the preset relationship recognition network and the intent type, and obtain the feature word association information corresponding to the target text information ;

An information query result acquisition unit, configured to query a prestored information database according to the intent type and the feature word association information, so as to obtain an information query result corresponding to the query request information.
A computer device, comprising a memory, a processor, and a computer program stored on the memory and operable on the processor, wherein the processor implements the following steps when executing the computer program:

If the input query request information is received, the corresponding target text information is extracted from the query request information;

Analyzing the target text information according to a preset text intent analysis model to obtain a corresponding intent type;

According to the preset relationship recognition network and the intent type, the association relationship between the feature words in the target text information is identified, and the feature word association information corresponding to the target text information is obtained;

A pre-stored information database is queried according to the intent type and the associated information of the characteristic words, so as to obtain an information query result corresponding to the query request information.
The computer device according to claim 9, wherein said extracting corresponding target text information from said query request information comprises:

judging the information type of the query request information;

If the information type of the query request information is a voice information type, then recognize the voice information contained in the query request information according to a pre-stored voice recognition model to obtain target text information corresponding to the query request information;

If the information type of the query request information is a text information type, the query request information is determined as target text information.
The computer device according to claim 10, wherein the speech recognition model includes a noise judgment rule and a text information acquisition model, and the speech information contained in the query request information is recognized according to the pre-stored speech recognition model to obtain a The target text information corresponding to the query request information includes:

judging whether the voice information in the query request information contains noise according to the noise judging rule;

If the voice information in the query request information does not contain noise, identify the voice information in the query request information according to the text information acquisition model, so as to obtain target text information corresponding to the query request information;

If the voice information in the query request information contains noise, feedback re-input prompt information to prompt re-input of the query request information.
The computer device according to claim 9, wherein the text intent analysis model includes word segmentation processing rules and a classification neural network, and the target text information is analyzed according to the preset text intent analysis model to obtain the corresponding intent type ,include:

performing word segmentation processing on the target text information according to the word segmentation processing rules to obtain word segmentation results;

Input the word segmentation result and the character feature vector into the classification neural network to obtain the intent type corresponding to the target text information.
The computer device according to claim 12, wherein the word segmentation processing rules include a vector conversion database and a part-of-speech tagging network, and performing word segmentation processing on the target text information according to the word segmentation processing rules to obtain word segmentation results includes:

Obtaining a corresponding character feature vector from the target text information according to the vector conversion database;

Carry out part-of-speech tagging to the character feature vector according to the part-of-speech tagging network, and obtain the part-of-speech tagging information corresponding to each character in the target text information;

The target text information is subjected to word segmentation processing according to the part-of-speech tagging information corresponding to each character, and a corresponding word segmentation result is obtained.
The computer device according to claim 9, wherein the querying of the pre-stored information database according to the type of intent and the associated information of the characteristic words to obtain an information query result corresponding to the query request information includes:

Generate a corresponding associated query statement according to a plurality of associated characteristic words in the characteristic word association information and the intent type;

Generate a separate query statement according to the feature words that do not have an association relationship in the feature word association relationship and the intent type;

According to the associated query statement and/or the individual query statement, the information database is queried to obtain a corresponding information query result.
The computer device according to claim 9, further comprising:

Upload the information query result to the block chain.
A computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following operations are performed:

If the input query request information is received, the corresponding target text information is extracted from the query request information;

Analyzing the target text information according to a preset text intent analysis model to obtain a corresponding intent type;

According to the preset relationship recognition network and the intent type, the association relationship between the feature words in the target text information is identified, and the feature word association information corresponding to the target text information is obtained;

A pre-stored information database is queried according to the intent type and the associated information of the characteristic words, so as to obtain an information query result corresponding to the query request information.
The computer-readable storage medium according to claim 16, wherein said extracting corresponding target text information from said query request information comprises:

judging the information type of the query request information;

If the information type of the query request information is a voice information type, then recognize the voice information contained in the query request information according to a pre-stored voice recognition model to obtain target text information corresponding to the query request information;

If the information type of the query request information is a text information type, the query request information is determined as target text information.
The computer-readable storage medium according to claim 17, wherein the speech recognition model includes a noise judgment rule and a text information acquisition model, and the speech information contained in the query request information is recognized according to the pre-stored speech recognition model To obtain the target text information corresponding to the query request information, including:

judging whether the voice information in the query request information contains noise according to the noise judging rule;

If the voice information in the query request information does not contain noise, identify the voice information in the query request information according to the text information acquisition model, so as to obtain target text information corresponding to the query request information;

If the voice information in the query request information contains noise, feedback re-input prompt information to prompt re-input of the query request information.
The computer-readable storage medium according to claim 16, wherein the text intent analysis model includes word segmentation processing rules and a classification neural network, and the target text information is analyzed according to the preset text intent analysis model to obtain corresponding intent types, including:

performing word segmentation processing on the target text information according to the word segmentation processing rules to obtain word segmentation results;

Input the word segmentation result and the character feature vector into the classification neural network to obtain the intent type corresponding to the target text information.
The computer-readable storage medium according to claim 19, wherein the word segmentation processing rules include a vector conversion database and a part-of-speech tagging network, and performing word segmentation processing on the target text information according to the word segmentation processing rules to obtain word segmentation results, include:

Obtaining a corresponding character feature vector from the target text information according to the vector conversion database;

Carry out part-of-speech tagging to the character feature vector according to the part-of-speech tagging network, and obtain the part-of-speech tagging information corresponding to each character in the target text information;

The target text information is subjected to word segmentation processing according to the part-of-speech tagging information corresponding to each character, and a corresponding word segmentation result is obtained.