Intelligent customer service interactive response method and system based on semantic analysis

Abstract

The invention relates to the technical field of semantic analysis, in particular to an intelligent customer service interactive response method and system based on semantic analysis, and the method comprises the steps: obtaining the historical data of an intelligent customer service and the voice data currently input by a user, and converting the voice data into a text statement; preprocessing the text statement and carrying out word segmentation; obtaining the average phrase length of the statement where each phrase is located, and obtaining the phrase rationality of each phrase; obtaining a continuous length mean value of phrases with the same part of speech in the statement in which each phrase is located, and obtaining the unpopular certainty degree of each phrase; obtaining a semantic reasonable value of each phrase; obtaining a segmentation threshold value of the semantic reasonable value, and judging whether each word group is interfered or not; evaluating the current text statement to obtain an accurate text statement; and finishing intelligent customer service interaction response. According to the method, the accuracy of semantic analysis is improved by obtaining the more accurate text statement of the user.

Description

Technical Field

[0001] This application relates to the technical field of semantic analysis, and specifically to an intelligent customer service interaction response method and system based on semantic analysis. Background Art

[0002] With the popularization of the Internet and mobile Internet, people's demand for customer service is also increasing. Traditional customer service methods have problems such as low efficiency and long response time, and it is difficult to meet the needs of users. The intelligent customer service system based on artificial intelligence can automatically answer users' questions and provide fast and accurate services, and has gradually become an important development direction in the customer service field.

[0003] Currently, the interaction response of intelligent customer service mainly uses speech recognition technology to convert the content in the user's speech into text, and then through text preprocessing technology, word segmentation of the text, removal of stop words, etc. are carried out, and combined with a deep learning model, semantic analysis and understanding of the user's question are realized, and dialogue management technology is used to achieve smooth multi-round conversations. However, the accuracy of semantic analysis and understanding is based on complete and clear user dialogue content. When performing text preprocessing, if accurate word segmentation cannot be performed on the user dialogue content text, it may cause incorrect semantic analysis and understanding, and ultimately lead to the inability to smoothly carry out the interaction of intelligent customer service. The patent "CN116881444A Text Preprocessing Method, Device, Medium and Equipment Based on Text Classification Model" mentions splitting the text to be preprocessed to obtain multiple sentences, then determining the sentences containing key information among all sentences, and finally sorting all sentences so that all sentences containing key information are closely arranged together, thereby reducing the probability of key information loss caused by text truncation. However, there are still certain defects in the above method. During the process of intelligent customer service interaction response, the environment during user dialogue varies greatly. During the process of text word segmentation, it may be affected by external environmental interference during user input (such as background voices, scene noises, etc., that is, the cocktail party effect), resulting in the recognized text being mixed with content not input by the user, and then it is impossible to accurately perform word segmentation during the word segmentation process, ultimately affecting the accuracy of semantic analysis. Therefore, there is an urgent need for a method that can solve the low accuracy of semantic analysis caused by inaccurate sentence word segmentation. Summary of the Invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide an intelligent customer service interaction response method and system based on semantic analysis, and the specific technical solutions adopted are as follows:

[0005] In the first aspect, an embodiment of this application provides an intelligent customer service interaction response method based on semantic analysis, and the method includes the following steps:

[0006] Obtain the historical data of the intelligent customer service and the current input voice data of the user, and convert the voice data into text statements; preprocess the text statements and perform word segmentation;

[0007] According to the total number of words and the total number of phrases in the sentence where each phrase is located, obtain the average phrase length of the sentence where each phrase is located, and combine the frequency of each phrase appearing in the historical data, as well as the frequency of the combination of each phrase and its adjacent phrases before and after appearing in the historical data, to obtain the phrase rationality of each phrase;

[0008] According to the number of phrases of the same part of speech continuously appearing in the sentence where each phrase is located, obtain the average continuous length of the phrases of the same part of speech in the sentence where each phrase is located, and combine the frequency of the adjacent phrases before and after each phrase appearing in the historical data, to obtain the cold certainty of each phrase; obtain the semantic rational value of each phrase according to the phrase rationality and cold certainty of each phrase; obtain the segmentation threshold of the semantic rational value, compare the semantic rational value of each phrase with the segmentation threshold, and determine whether each phrase is interfered;

[0009] Evaluate the current text statement according to the proportion of the interfered phrases in the current text statement, obtain the accurate text statement; complete the intelligent customer service interactive response according to the accurate text statement.

[0010] Preferably, the average phrase length of the sentence where each phrase is located is the ratio of the total number of words in the sentence where each phrase is located to the total number of phrases.

[0011] Preferably, the calculation formula for the phrase rationality of each phrase is: A i = B i + C i + V i ; In the formula, A i is the phrase rationality of the i-th phrase, B i is the frequency of the i-th phrase appearing in the historical data, C i is the frequency of the combination of the i-th phrase and its adjacent phrases before and after appearing in the historical data, V i is the average phrase length of the sentence where the i-th phrase is located.

[0012] Preferably, the method for determining the average continuous length of the phrases of the same part of speech in the sentence where each phrase is located is: when there are two or more continuously arranged phrases of the same part of speech in the sentence where each phrase is located, then count the number of phrases in each continuous phrase of the same part of speech, and calculate the average value of all the above-mentioned phrase numbers in the sentence where each phrase is located as the average continuous length of the phrases of the same part of speech in the sentence where each phrase is located.

[0013] Preferably, the calculation formula for the cold certainty of each phrase is: In the formula, D iis the cold start confidence of the i-th phrase, E i is the sum of the occurrence frequencies of the adjacent phrases before and after the i-th phrase in the historical data, F i is the average value of the continuous length of the phrases with the same part of speech in the sentence where the i-th phrase is located, and τ is a preset parameter.

[0014] Preferably, the semantic reasonable value of each phrase is the positive fusion result of the phrase rationality and the cold start confidence of each phrase.

[0015] Preferably, the specific process of determining whether each phrase is interfered is as follows: when the semantic reasonable value of the phrase is greater than or equal to the segmentation threshold, the phrase is not interfered; otherwise, the phrase is interfered.

[0016] Preferably, the specific process of evaluating the current text sentence to obtain an accurate text sentence is as follows: if the proportion of the phrases with semantic reasonable values greater than or equal to the segmentation threshold in the current sentence exceeds the preset threshold, it means that the content of the current text sentence can accurately express what the user wants to express, then the current text sentence is an accurate text sentence; otherwise, it means that the content of the current text sentence cannot accurately express what the user wants to express, then the user needs to re-enter to obtain an accurate text sentence.

[0017] Preferably, the process of completing the intelligent customer service interactive response according to the accurate text sentence is as follows: after obtaining the accurate text sentence, use a semantic analysis model to perform semantic analysis on the text sentence to obtain semantic information data, and match it with the content in the knowledge base; retrieve the corresponding answer in the Q&A knowledge base according to the matching result, and perform content rendering on the answer to obtain the response content, and use the response content as the output of the intelligent customer service.

[0018] In a second aspect, the embodiment of the present application further provides an intelligent customer service interactive response system based on semantic analysis, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the above-mentioned intelligent customer service interactive response method based on semantic analysis.

[0019] The present application has at least the following beneficial effects:

[0020] By analyzing the characteristics of phrases in text sentences affected by the cocktail party effect, this application constructs the semantic reasonable value of each sentence, which can determine whether there is interference in each phrase and whether the semantics expressed by the current text sentence is accurate, and re-obtains accurate text sentences when the text sentence is inaccurate. It solves the problem of deviation in the semantic analysis and recognition process caused by the inability to accurately segment the text sentences input by users during the intelligent customer service interaction response process, improves the accuracy of semantic analysis, and provides more accurate decision support for the intelligent customer service interaction response. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] To more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0022] Figure 1 It is a flowchart of the steps of an intelligent customer service interaction response method based on semantic analysis provided by an embodiment of the present application;

[0023] Figure 2 It is a processing flowchart for processing the obtained text sentences provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] In order to further elaborate on the technical means and effects adopted by the present application to achieve the intended invention purpose, the following will, in conjunction with the drawings and preferred embodiments, detail the specific implementation manners, structures, features, and effects of the intelligent customer service interaction response method and system based on semantic analysis proposed according to the present application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs.

[0026] The following will specifically describe the specific solutions of the intelligent customer service interaction response method and system based on semantic analysis provided by the present application in conjunction with the drawings.

[0027] Please refer to Figure 1 , which shows a flowchart of the steps of an intelligent customer service interaction response method based on semantic analysis provided by an embodiment of the present application. The method includes the following steps:

[0028] Step 1: Obtain the historical data of the intelligent customer service and the current voice data input by the user, and convert the voice data into a text statement; preprocess the text statement and perform word segmentation.

[0029] Obtain the historical data through the intelligent customer service interactive response system and receive the voice input data from the user. The system first uses automatic speech recognition (ASR) technology to convert the received voice data of the user's input into text data, and then uses regular expressions to remove the punctuation marks in the text. Then, word segmentation is performed on the text. The word segmentation algorithm can be, but is not limited to, the forward maximum matching method, the backward maximum matching method, and the minimum segmentation method. This embodiment does not make a limitation.

[0030] Step 2: According to the total number of words and the total number of phrases in the sentence where each phrase is located, obtain the average phrase length of the sentence where each phrase is located, and combine the frequency of each phrase appearing in the historical data, as well as the frequency of the combination of each phrase and the adjacent phrases before and after in the historical data, to obtain the phrase rationality of each phrase.

[0031] For the voice input of the user, the other voices in the background are one of the important interference factors, which will ultimately lead to the situation that in the recognized text statement, the user's input is mixed with the background voice content, and the smoothness of the recognized text statement decreases (i.e., the "cocktail party effect"). For example, if the user wants to input "I need to modify user information", and the background voice is "It takes an hour to take a bus here to the central square", the finally recognized voice content after the content is mixed may be "I need to take a bus to modify user information slightly", which further leads to the inability to accurately segment "carry out" and "modify" during word segmentation, making it impossible to recognize the true input semantics of the user during subsequent semantic analysis.

[0032] For intelligent customer services in the same field (such as intelligent customer services dedicated to e-commerce, banking, and video software), the user input content received is usually relatively similar. For example, for an intelligent customer service for e-commerce, the user may usually input "I need a refund", "When will it be shipped", "I need to exchange goods". Therefore, for intelligent customer services in the same field, the probability of the correct combination of phrases appearing is more frequent, such as the above "need", "refund", etc., and the probability of the combination of two words or phrases appearing is frequent, such as "I" and "need", "what" and "time", etc. Therefore, if the cocktail party effect occurs in a sentence of the user's input, some combinations may appear with a very low frequency in the historical data after word segmentation.

[0033] In addition, for normal sentences, the number of phrases after word segmentation is limited. However, when a sentence is affected by the cocktail party effect, it may cause the phrases in the sentence to be split, and as a result, more single words will appear during word segmentation. For example, the sentence "I need to modify user information" mentioned above is normally segmented as: I / need / to / modify / user information. After being affected, the sentence becomes "I need to enter and modify user information", and after segmentation, it may be: I / need / to / enter / and / modify / user information. Therefore, when a sentence is affected, the average phrase length after word segmentation will be shorter, and vice versa.

[0034] As a preferred embodiment, according to the total number of words and the total number of phrases in the sentence where each phrase is located, obtain the average phrase length of the sentence where each phrase is located. Combine the frequency of each phrase appearing in the historical data, and the frequency of the combination of each phrase and its adjacent phrases before and after appearing in the historical data, to obtain the phrase rationality of each phrase.

[0035] In this embodiment, the phrase rationality of the i-th phrase is denoted as A i , and its specific expression is: A i = B i + C i + V i ; In the formula, A i is the phrase rationality of the i-th phrase, B i is the frequency of the i-th phrase appearing in the historical data, C i is the frequency of the combination of the i-th phrase and its adjacent phrases before and after appearing in the historical data, V i is the average phrase length of the sentence where the i-th phrase is located. The average phrase length refers to the ratio of the total number of words in the sentence where the i-th phrase is located to the total number of phrases.

[0036] A i The larger the value of A

[0037] Step 3: According to the number of phrases of the same part of speech continuously appearing in the sentence where each phrase is located, obtain the average continuous length of the phrases of the same part of speech in the sentence where each phrase is located. Combine the frequency of the adjacent phrases before and after each phrase appearing in the historical data to obtain the cold certainty of each phrase; according to the phrase rationality and cold certainty of each phrase, obtain the semantic rational value of each phrase; obtain the segmentation threshold of the semantic rational value, and compare the semantic rational value of each phrase with the segmentation threshold to determine whether each phrase is interfered.

[0038] Even though intelligent customer service may face a large number of user inputs every day, there will be some unpopular content in the intelligent customer service responses in each field. Compared with popular content, the number of user inputs is less, but it does not mean that there is no user input. When user input occurs, if only the above method is used for calculation, the unpopular phrases may be judged as disturbed phrases, and then their semantic expressions may be judged as inaccurate, leading to misjudgment of user input. Therefore, further analysis is needed for this situation.

[0039] For unpopular phrase combinations, even if the probability of their occurrence in historical data is low, the overall semantics of the sentence in which the phrase is located is accurate without interference, that is, the subject-predicate-object relationship of each participle and the probability of occurrence of non-unpopular phrases in the sentence are high. For example, in "I repair the oblique detonation engine", "I" and "repair" are both normal high-frequency phrases, and the subject-predicate-object relationship of "I", "repair", and "oblique detonation engine" in the sentence is a normal subject, predicate, and object. When the cocktail party effect occurs, the phrases in the sentence will be disrupted, which will lead to changes in the subject-predicate-object structure of the sentence, and multiple verbs or nouns may be superimposed.

[0040] In summary, by analyzing the independent occurrence frequency of the phrases before and after the phrase and the length of the phrases with the same part of speech in the sentence where the phrase is located, the unpopular confidence of each phrase is constructed to characterize the possibility that each phrase belongs to an undisturbed unpopular phrase. In this embodiment, the unpopular confidence of the i-th phrase is denoted as D i , its specific expression is: Where D i is the unpopularity confidence of the i-th phrase, E i is the sum of the frequencies of the adjacent phrases before and after the i-th phrase in the historical data, F i is the mean continuous length of the phrases with the same part of speech in the sentence where the ith phrase is located, τ is a preset parameter to prevent the denominator from being 0, and 0.1 is taken in this embodiment. Among them, the calculation method of the mean continuous length of the phrases with the same part of speech in the sentence where the ith phrase is located is: when two or more consecutive phrases with the same part of speech appear in the sentence where the ith phrase is located, the number of phrases in each segment of consecutive phrases with the same part of speech is counted, and the mean of the number of all the phrases in the sentence is calculated as the mean continuous length of the phrases with the same part of speech in the sentence. It should be noted that in this scheme, the knowledge base content is used to identify the part of speech of the phrase in a rule-based manner, and the implementer can use open AI platform, deep learning and other methods to identify the part of speech according to actual efficiency requirements.

[0041] When the underdog is certain iWhen is larger, the frequency of the adjacent phrases of the i-th phrase in the historical data is greater, and the mean length of the consecutive appearances of the same part-of-speech phrases in the sentence where the phrase is located is smaller, which means that the phrase is more likely to be an undisturbed unpopular phrase.

[0042] Furthermore, as a preferred implementation, according to the forward fusion results of the phrase rationality and unpopular confidence of each phrase, a semantic rationality value of each phrase is constructed to characterize the possibility of interference of each phrase.

[0043] In this embodiment, the semantic reasonableness value of the i-th phrase is recorded as G i , its specific expression is: G i =A i +D i Where G i is the semantically reasonable value of the i-th phrase, A i is the phrase rationality of the i-th phrase, D i is the unpopularity confidence of the ith phrase. When the phrase rationality of the ith phrase is higher and the unpopularity confidence is higher, it means that the phrase is more likely to be an undisturbed phrase in the sentence and the semantics expressed is more reasonable.

[0044] In another embodiment, the semantic reasonableness value of the i-th phrase can be calculated as follows: i =A i ×D i Where G i is the semantically reasonable value of the i-th phrase, A i is the phrase rationality of the i-th phrase, D i is the unpopularity confidence of the i-th phrase.

[0045] The semantically reasonable values of all phrases are taken as input, and the segmentation threshold of the semantically reasonable value is output by cross-validation. When the semantically reasonable value of a phrase is greater than or equal to the segmentation threshold, the phrase is not interfered with and the semantics expressed is more accurate. Otherwise, the phrase is interfered with and the semantics expressed is less accurate.

[0046] Step 4: Evaluate the current text sentence based on the proportion of the disturbed phrases in the current text sentence to obtain an accurate text sentence; complete the intelligent customer service interactive response based on the accurate text sentence.

[0047] Through the above steps, the semantic rationality of each phrase in the text statement of the current user can be determined. Further, the text statement of the current user can be evaluated. The specific process is as follows: When the proportion of phrases with a semantic rationality value greater than or equal to the segmentation threshold in the text statement of the current user exceeds H%, it indicates that the content of the current text statement can accurately express what the user wants to express. Then the current text statement is an accurate text statement; otherwise, it indicates that the content of the current text statement cannot accurately express what the user wants to express. Then the user needs to re-enter to obtain an accurate text statement. It should be noted that in this embodiment, H is taken as 95, and the implementer should adjust it according to the actual semantic analysis accuracy requirements. The processing flow chart for processing the obtained text statement is as Figure 2 shown.

[0048] In this embodiment, the BERT (Bidirectional Encoder Representations from Transformers) model is used to perform semantic analysis and recognition on the above-obtained accurate text statement to obtain semantic information data. The implementer can also use other semantic analysis models for analysis and recognition. The intelligent customer service matches the semantic information data with the relevant content in the knowledge base, retrieves the corresponding answer in the Q&A knowledge base according to the matching result, and finally renders the content of the retrieved answer to obtain the final response content, and uses the response content as the output of the intelligent customer service to complete the interactive response of the intelligent customer service.

[0049] Thus, an intelligent customer service interactive response method and system based on semantic analysis are completed.

[0050] Based on the same inventive concept as the above method, an embodiment of the present application also provides an intelligent customer service interactive response system based on semantic analysis, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above methods of the intelligent customer service interactive response method based on semantic analysis.

[0051] It should be noted that: The above sequence of embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above describes specific embodiments of this specification. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0052] Each embodiment in this specification is described in a progressive manner. The same or similar parts between each embodiment can be referred to each other, and the key points of each embodiment are the differences from other embodiments.

[0053] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the principle of the present application shall be included within the protection scope of the present application.

Claims

1. An intelligent customer service interaction response method based on semantic analysis, characterized in that, The method includes the following steps: Obtain the historical data of the intelligent customer service and the voice data currently input by the user, and convert the voice data into a text statement; preprocess the text statement and perform word segmentation; According to the total number of words and the total number of phrases in the statement where each phrase is located, obtain the average phrase length of the statement where each phrase is located, and combine the frequency of each phrase appearing in the historical data, as well as the frequency of the combination of each phrase and its adjacent phrases before and after appearing in the historical data, to obtain the phrase rationality of each phrase; According to the number of phrases of the same part of speech continuously appearing in the statement where each phrase is located, obtain the average continuous length of the phrases of the same part of speech in the statement where each phrase is located, and combine the frequency of the adjacent phrases before and after each phrase appearing in the historical data, to obtain the cold certainty of each phrase; obtain the semantic rational value of each phrase according to the phrase rationality and the cold certainty of each phrase; obtain the segmentation threshold of the semantic rational value, compare the semantic rational value of each phrase with the segmentation threshold, and determine whether each phrase is interfered; Evaluate the current text statement according to the proportion of the interfered phrases in the current text statement, and obtain an accurate text statement; complete the intelligent customer service interactive response according to the accurate text statement.

2. The intelligent customer service interaction response method based on semantic analysis according to claim 1, characterized in that The average phrase length of the statement where each phrase is located is the ratio of the total number of words in the statement where each phrase is located to the total number of phrases.

3. The intelligent customer service interaction response method based on semantic analysis according to claim 1, characterized in that, The formula for calculating the phrase rationality of each phrase is: A i = B i + C i + V i ; where A i is the phrase rationality of the i-th phrase, B i is the frequency of the i-th phrase appearing in the historical data, C i is the frequency of the combination of the i-th phrase and the adjacent phrases before and after appearing in the historical data, and V i is the average phrase length of the sentence where the i-th phrase is located.

4. The intelligent customer service interaction response method based on semantic analysis according to claim 1, wherein The method for determining the average continuous length of the phrases of the same part of speech in the statement where each phrase is located is: when there are two or more continuously arranged phrases of the same part of speech in the statement where each phrase is located, then count the number of phrases in each continuous segment of phrases of the same part of speech, and calculate the average value of all the said phrase numbers in the statement where each phrase is located as the average continuous length of the phrases of the same part of speech in the statement where each phrase is located.

5. The intelligent customer service interactive response method based on semantic analysis according to claim 1, characterized in that The calculation formula for the cold popularity confidence of each phrase is as follows: In the formula, D i is the cold popularity confidence of the i-th phrase, and E i is the sum of the occurrence frequencies of the adjacent phrases before and after the i-th phrase in the historical data, and F i is the average value of the consecutive lengths of the phrases with the same part of speech in the sentence where the i-th phrase is located, and τ is a preset parameter.

6. The intelligent customer service interaction response method based on semantic analysis according to claim 1, characterized in that, The semantic rational value of each phrase is the positive fusion result of the phrase rationality and the cold certainty of each phrase.

7. The intelligent customer service interaction response method based on semantic analysis according to claim 1, wherein The specific process of determining whether each phrase is interfered is: when the semantic rational value of the phrase is greater than or equal to the segmentation threshold, then the phrase is not interfered; otherwise, the phrase is interfered.

8. The intelligent customer service interaction response method based on semantic analysis according to claim 1, characterized in that, The specific process of evaluating the current text statement and obtaining an accurate text statement is: if the proportion of the phrases with semantic rational values greater than or equal to the segmentation threshold in the current statement exceeds the preset threshold, it means that the content of the current text statement can accurately express what the user wants to express, then the current text statement is the accurate text statement; otherwise, it means that the content of the current text statement cannot accurately express what the user wants to express, then the user needs to re-enter to obtain an accurate text statement.

9. The intelligent customer service interaction response method based on semantic analysis according to claim 1, wherein The process of completing the intelligent customer service interactive response according to the accurate text statement is: after obtaining the accurate text statement, use a semantic analysis model to perform semantic analysis on the text statement to obtain semantic information data, and match it with the content in the knowledge base; retrieve the corresponding answer in the Q&A knowledge base according to the matching result, and perform content rendering on the answer to obtain the response content, and use the response content as the output of the intelligent customer service.

10. An intelligent customer service interactive response system based on semantic analysis, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent customer service interactive response method based on semantic analysis according to any one of claims 1-9.

Images