TWI807569B - Auto-reply system and auto-reply method based on instant message service - Google Patents

Abstract

An auto-reply system and an auto-reply method based on an instant message service (IMS) are provided. The auto-reply method includes: obtaining a question and answer database and a negative emotion database, wherein the question and answer database includes a first question message and a first answer message corresponding to the first question message, wherein the negative emotion database includes a first word; obtaining a first IMS message from a first terminal device; replying a second IMS message including the first answer message to the first terminal device in response to the first IMS message matching with the first question message; obtaining a third IMS message corresponding to the second IMS message from the first terminal device; transmitting an inform message to a second terminal device in response to the third IMS message matching with the first word.

Description

Automatic answering system and automatic answering method based on instant messaging service

The invention relates to an automatic answering system and an automatic answering method based on an instant messaging service.

The rise of automated response systems (or chat bots) has a lot to do with changing user habits. In recent years, the service with the highest percentage of usage among mobile applications is instant messaging. Compared with requiring users to reinstall software or use website services, existing mobile instant messaging obviously provides more intuitive application scenarios and operation modes. Compared with using telephone or E-mail for customer service, instant messaging service can provide faster processing speed and improve customer satisfaction.

In the traditional maintenance and operation mode of customer service based on instant messaging, most instant messaging still relies on the handling of real customer service. However, most of the maintenance and operation colleagues face repeated or highly relevant problems from different customers every time. In the case of lack of manpower for maintenance and operation, use dialogue robots to answer customer questions first, and only when necessary, real people will take over. In addition to reducing the cost of customer service manpower, it can also provide 24-hour question-and-answer service at the same time. In turn, unnecessary human and time resource consumption can be reduced, and customer satisfaction can be improved.

However, conversational bots are not necessarily effective at solving problems customers encounter. In this case, if the customer cannot be handed over to a real person in real time, the service experience of the customer may be affected.

The invention provides an automatic answering system and an automatic answering method based on an instant messaging service, which can automatically respond to a customer's question, and can transfer the customer to a customer service personnel for processing when necessary.

An automatic answering system based on the instant messaging service of the present invention includes a processor, a storage medium and a transceiver. The transceiver is communicatively connected to the first terminal device and the second terminal device. The storage medium stores multiple modules. The processor is coupled to the storage medium and the transceiver, and accesses and executes multiple modules, wherein the multiple modules include a negative emotion lexicon, a question and answer lexicon, a message receiving module, a question and answer retrieval module, and a customer service processing module. The negative emotion thesaurus stores the first vocabulary. The question-and-answer thesaurus stores the first question message and the first answer message corresponding to the first question message. The message receiving module obtains the first instant messaging message from the first terminal device through the transceiver. In response to the match between the first instant messaging message and the first question message, the question and answer retrieval module replies a second instant messaging message including the first answer message to the first terminal device through the transceiver. The customer service processing module obtains the third instant messaging message corresponding to the second instant messaging message from the first terminal device through the transceiver, and responds to the third instant messaging message matching the first vocabulary configured to transmit the notification message to the second terminal device through the transceiver.

In an embodiment of the present invention, the above-mentioned customer service processing module establishes a communication channel between the first terminal device and the second terminal device in response to the matching of the third instant messaging message with the first vocabulary.

In an embodiment of the present invention, the above-mentioned question and answer lexicon further includes a second question message, wherein the question and answer retrieval module calculates a first similarity between the first instant messaging message and the first question message and a second similarity between the first instant messaging message and the second question message, wherein the question and answer retrieval module determines that the first instant messaging message matches the first question message in response to the first similarity being greater than the second similarity.

In an embodiment of the present invention, the question-answer thesaurus further includes a second answer message corresponding to the first question message, wherein the question-answer retrieval module calculates a first similarity between the first instant messaging message and the first answer message and a second similarity between the first instant messaging message and the second answer message, wherein the question-answer retrieval module replies the first answer message to the first terminal device in response to the first similarity being greater than the second similarity.

In an embodiment of the present invention, the above question and answer retrieval module calculates the first similarity based on the best matching 25 algorithm.

In an embodiment of the present invention, the above question and answer retrieval module marks the second vocabulary in the first question message with a part of speech and a bag of words index based on natural language processing, wherein the question and answer retrieval module calculates the first similarity according to the part of speech and the bag of words index.

In an embodiment of the present invention, the above question and answer retrieval module pre-stores a reference dictionary, and responds to the second word in the first question message and the first word in the reference dictionary The three words are matched and the part of speech and the bag of words index are marked for the second word according to the third word, wherein the question and answer retrieval module calculates the first similarity according to the part of speech and the bag of words index.

In an embodiment of the present invention, the above-mentioned modules further include a data processing module. The data processing module obtains the file containing the first question message and the first answer message through the transceiver, wherein the data processing module generates a mapping relationship between the first question message and the first answer message according to the file, and builds a question-and-answer lexicon according to the mapping relationship.

In an embodiment of the present invention, the above-mentioned customer service processing module obtains the file through the transceiver, and marks the emotional state for the first word in the file based on natural language processing, wherein the customer service processing module adds the first word into the negative emotion lexicon in response to the emotional state being a negative emotional state.

In an embodiment of the present invention, the above-mentioned customer service processing module increases a count value in response to the third instant messaging message matching the first vocabulary, and sends a notification message in response to the count value being greater than a threshold.

An automatic answering method based on an instant messaging service of the present invention, comprising: obtaining a question-and-answer lexicon and a negative emotion lexicon, wherein the question-answer lexicon includes a first question message and a first answer message corresponding to the first question message, wherein the negative emotion lexicon includes a first vocabulary; obtaining a first instant messaging message from a first terminal device; responding to a match between the first instant message message and the first question message, replying a second instant message message including the first answer message to the first terminal device; obtaining a third instant message message corresponding to the second instant message message from the first terminal device; A notification message is sent to the second terminal device when the third instant messaging message matches the first word.

Based on the above, the automatic answering system of the present invention can information to select the most appropriate answer to respond to the customer. When the customer is not satisfied with the answer from the automatic response system, the automatic response system can detect whether the customer has negative emotional reactions based on the customer's response, and transfer the customer's terminal device to the terminal device of the customer service personnel when the customer has a negative emotional reaction, so that the customer service personnel can serve the customer instead, so that the customer can get a better service experience.

10: Automatic response system

100: Processor

20: The first terminal device

200: storage media

210: Data processing module

220:Question and answer retrieval module

230: Message receiving module

240:Customer service processing module

250: Graphical User Interface

251, 252: dialog box

253, 254, 255: buttons

260: Question and Answer Thesaurus

270: Thesaurus of Negative Emotions

300: Transceiver

S301, S302, S303, S304, S401, S402, S403, S404, S501, S502, S503, S504, S601, S602, S603, S604, S605, S606, S701, S702, S703, S704, S801, S802, S803, S804, S805: steps

FIG. 1 shows a schematic diagram of an instant messaging service-based automatic answering system according to an embodiment of the present invention.

FIG. 2 is a schematic diagram of a first terminal device according to an embodiment of the present invention.

FIG. 3 shows a flow chart of building a question-answer thesaurus according to an embodiment of the present invention.

FIG. 4 is a flow chart of labeling vocabulary according to an embodiment of the present invention.

FIG. 5 shows a flow chart of establishing a negative emotion lexicon according to an embodiment of the present invention.

FIG. 6 shows a flow chart of determining an answer message according to an embodiment of the present invention.

FIG. 7 shows a flow chart of transferring customer service personnel according to an embodiment of the present invention.

FIG. 8 is a flow chart of an instant messaging service-based automatic answering method according to an embodiment of the present invention.

Fig. 1 shows a service based on instant messaging according to an embodiment of the present invention A schematic diagram of the automated answering system 10. The automatic answering system 10 can implement the function of a question answering robot (FAQ bot) to automatically respond to questions and messages raised by customers. The automatic answering system 10 can also detect the customer's emotional reaction according to the customer's reply, so as to transfer the customer to the customer service personnel in a timely manner when the customer has negative emotions. The automatic answering system 10 may include a processor 100 , a storage medium 200 and a transceiver 300 .

The processor 100 is, for example, a central processing unit (CPU), or other programmable general-purpose or special-purpose micro control unit (micro control unit, MCU), microprocessor (microprocessor), digital signal processor (digital signal processor, DSP), programmable controller, application specific integrated circuit (application specific integrated circuit, ASIC), graphics processor (graphics processing unit, GPU), image signal processor (image signal processor, ISP), image processing unit (image processing unit, IPU), arithmetic logic unit (arithmetic logic unit, ALU), complex programmable logic device (complex programmable logic device, CPLD), field programmable logic gate array (field programmable gate array, FPGA) or Other similar elements or combinations of the above elements. The processor 100 is connected to the storage medium 200 and the transceiver 300 , and accesses and executes a plurality of modules and various application programs stored in the storage medium 200 .

The storage medium 200 is, for example, any type of fixed or removable random access memory (random access memory, RAM), read-only memory (read-only memory, ROM), flash memory (flash memory), hard disk drive (hard disk drive, HDD), solid state drive (solid state drive, SSD) or Similar elements or a combination of the above elements are used to store multiple modules or various application programs that can be executed by the processor 100 . In this embodiment, the storage medium 200 can store multiple modules including a data processing module 210, a question and answer retrieval module 220, a message receiving module 230, a customer service processing module 240, a graphical user interface (graphical user interface, GUI) 250, a question and answer lexicon 260, and a negative emotion lexicon 270. The functions thereof will be described later.

The transceiver 300 transmits and receives signals in a wireless or wired manner. The transceiver 300 may also perform operations such as low noise amplification, impedance matching, frequency mixing, up or down frequency conversion, filtering, amplification, and the like.

The automatic answering system 10 can communicate with the customer's terminal device (or called "first terminal device") or the customer service personnel's terminal device (or called "second terminal device") through the transceiver 300 . The automatic answering system 10 can provide a graphical user interface 250 to the first terminal device, so as to provide customers with a question-and-answer robot service. FIG. 2 is a schematic diagram of a first terminal device 20 according to an embodiment of the present invention. The automatic answering system 10 can transmit the GUI 250 to the first terminal device 20 through the transceiver 300 . The GUI 250 may include a dialog box 251 for displaying instant messaging messages from customers and a dialog box 252 for displaying instant messaging messages responded by the robot service provided by the automatic answering system 10 . The GUI 250 may further include a button 253 for calling a robot (ie, the automatic answering system 10 ), a button 254 for calling a customer service agent, and a button 255 for searching files.

In one embodiment, the GUI 250 can start to provide the question answering robot service to the first terminal device 20 after the button 253 is pressed. for example, The automatic answering system 10 can allow the first terminal device 20 to send an instant messaging message to the automatic answering system 10 after the button 253 is pressed.

In one embodiment, the automatic answering system 10 can establish a communication channel between the first terminal device 20 and the second terminal device after the button 254 is pressed. In other words, the automatic answering system 10 can transfer the user of the first terminal device 20 (ie, the customer) to the user of the second terminal device (ie, the customer service personnel) after the button 254 is pressed.

In one embodiment, the automatic answering system 10 can access the file database outside the system through the transceiver 300 . After the automatic answering system 10 can press the button 255 , the automatic answering system 10 can establish a communication channel between the first terminal device 20 and the file database. The client can operate the first terminal device 20 to search and read the files in the file database.

The question-and-answer dictionary 260 can store question messages and answer messages corresponding to the question messages. In addition, the question-answer lexicon 260 can further record the part of speech (for example: noun, pronoun, adjective, verb, adverb, preposition or linking word, etc.) or bag of words (bag of words) index of the vocabulary in the information stored in the question-answer lexicon 260 (ie: question message or answer message). In one embodiment, one question message may correspond to multiple answer messages. For example, the answer message corresponding to the question message "is the customer service staff busy?" may include multiple answer messages such as "the customer service staff is busy" and "the customer service staff can provide services immediately".

The automatic answering system 10 can automatically generate the information in the question-and-answer thesaurus 260 according to the documents. FIG. 3 illustrates the flow of building a question-and-answer lexicon 260 according to an embodiment of the present invention map. In step S301 , the data processing module 210 can obtain a file including question information and answer information through the transceiver 300 . For example, the file is, for example, a dialogue record or a dialogue log between a customer service personnel and a customer in the past, wherein the dialogue record includes the question message raised by the customer and the answer message responded by the customer service personnel.

In step S302, the data processing module 210 may perform pre-processing on the file. Pre-processing may include but not limited to format judgment or content filtering. For example, the data processing module 210 can determine whether the file is unstructured data. If the file is unstructured data, the data processing module 210 can perform content filtering on the file, so as to delete unimportant or repeated fragmentary content. For example, the data processing module 210 can filter content such as blanks, page breaks, line breaks, specific fonts, main titles, subtitles, specific styles, pictures, or web links in the document.

In step S303, the data processing module 210 can establish a mapping relationship between question information and answer information. Specifically, the data processing module 210 can generate a mapping relationship of key-value pairs from unstructured files (or filtered files) through data cleaning. In one embodiment, the data processing module 210 can generate the key-value pair according to the format of the file, but the present invention does not limit the method of generating the key-value pair. For example, the data processing module 210 can set the content of the paragraph without a period in the file as a key, and set the content with a number sorted after the key as a value corresponding to the key, thereby generating a key-value pair. The data processing module 210 can define the key in the key-value pair (ie: the main title in the file) as the question message, and define the value in the key-value pair (ie: the content corresponding to the main title in the file) as the answer message. Since one key can correspond to multiple values, one question message can correspond to multiple answer messages.

In step S304 , the data processing module 210 can establish or update the question-answer thesaurus 260 according to the mapping relationship. The data processing module 210 can use JSON (JaveScript object notation) format to store the question message, one or more answer messages corresponding to the question message, and the mapping relationship between the question message and the answer message in the question-answer lexicon 260 .

The automatic answering system 10 can further mark part-of-speech and bag-of-words index for the vocabulary included in the message in the question-answer thesaurus 260 based on natural language processing (NLP). FIG. 4 is a flow chart of labeling vocabulary according to an embodiment of the present invention. In step S401 , the question and answer retrieval module 220 can normalize the information (for example, question information or answer information) in the question and answer thesaurus 260 . For example, the question-and-answer retrieval module 220 can filter specific tags, specific symbols, users, stop words or consecutively repeated words in the message.

In step S402, the question-answer retrieval module 220 may perform tokenization processing on the message, so as to divide the message into one or more words. Specifically, assuming that the message is a Chinese message, after completing the normalization of the message, the question and answer retrieval module 220 may segment the normalized message based on the Hidden Markov Model tokenizer, so as to obtain one or more words from the message. On the other hand, the question-and-answer retrieval module 220 can also segment the message according to the stop words in the message.

In step S403 , the question-answer retrieval module 220 can tag the vocabulary with part of speech and bag-of-words index. The question-and-answer retrieval module 220 can mark part-of-speech and bag-of-words index for vocabulary based on natural language processing.

Due to the poor effect of traditional Chinese word breakers, in order to more correctly To summarize the parts of speech, the question and answer retrieval module 220 can pre-store a reference dictionary, wherein the reference dictionary can include multiple words and multiple parts of speech respectively corresponding to the multiple words. When the vocabulary contained in the message in the question-answer thesaurus 260 matches the vocabulary in the reference dictionary, the question-answer retrieval module 220 can tag the vocabulary in the message according to the part of speech of the vocabulary in the reference dictionary. For example, suppose the reference dictionary contains "customer service personnel (noun)". If the message in the question-and-answer dictionary 260 is "is the customer service personnel busy?", the question-and-answer retrieval module 220 may mark the word "customer service personnel" in the message "is the customer service personnel busy?" in response to the word "customer service personnel" in the message "is the customer service personnel busy?" matching the word "customer service personnel (noun)" in the reference dictionary.

On the other hand, the question-and-answer retrieval module 220 can annotate the bag-of-words index for vocabulary to find important words based on the bag-of-words model. The question and answer retrieval module 220 is also based on the term frequency-inverse document frequency (term frequency-inverse document frequency, tf-idf) statistical search algorithm to analyze the importance of a word in the message in the question and answer thesaurus 260, so the question and answer retrieval module 220 can find out the frequency and important words of a certain word in the question and answer thesaurus 260 through the bag of words and tf-idf.

In step S404, the question-and-answer retrieval module 220 updates the question-and-answer thesaurus 260 according to the vocabulary marked with the word-of-speech and the bag-of-words index, so that the question-and-answer thesaurus 260 further stores the part-of-speech and the bag-of-words index of the vocabulary in each message (for example: question message or answer message) in addition to storing the question message, the answer message, and the mapping relationship.

The negative emotion lexicon 270 can store a plurality of words corresponding to negative emotion states (eg, anger or impatience). The customer service processing module 240 can automatically Generate the data in the negative emotion lexicon 270 accurately. FIG. 5 shows a flow chart of establishing a negative emotion lexicon 270 according to an embodiment of the present invention. In step S501 , the customer service processing module 240 can obtain the file through the transceiver 300 . For example, the file is, for example, a record or a log of conversations between customer service personnel and customers in the past.

In step S502, the customer service processing module 240 may perform pre-processing on the file. Pre-processing may include but not limited to format judgment or content filtering. For example, the customer service processing module 240 can determine whether the file is unstructured data. If the file is unstructured data, the customer service processing module 240 can perform content filtering on the file to delete unimportant or repeated fragmentary content. For example, the customer service processing module 240 can filter content such as blanks, page breaks, line breaks, specific fonts, main titles, subtitles, specific styles, pictures or web links in the file. In one embodiment, each vocabulary in the document has been tagged with part of speech and bag-of-words index.

In step S503, the customer service processing module 240 can mark the emotional state for the vocabulary in the file based on natural language processing. The customer service processing module 240 can judge the emotional state of the vocabulary in the document according to a sentiment analysis algorithm, and mark the emotional state for the vocabulary according to the judgment result.

In step S504 , the customer service processing module 240 may add the vocabulary into the negative emotion thesaurus 270 in response to the emotional state of the vocabulary being a negative emotional state, so as to establish or update the negative emotion thesaurus 270 .

After the automatic answering system 10 receives the question message from the first terminal device 20, the automatic answering system 10 can automatically generate and reply an answer message corresponding to the question message. Fig. 6 shows the process of determining the answer message according to an embodiment of the present invention picture. In step S601 , the message receiving module 230 can obtain a first instant messaging message from the first terminal device 20 through the transceiver 300 , wherein the first instant messaging message can include a question that the customer wants to ask.

In step S602, the question and answer retrieval module 220 may perform pre-processing on the first instant messaging message. Pre-processing may include but not limited to format judgment or content filtering. For example, the question-and-answer retrieval module 220 may perform content filtering on the first instant messaging message, and delete unimportant or repeated fragmentary content.

In step S603 , the question-answer retrieval module 220 can mark the vocabulary in the first instant messaging message with part of speech and bag-of-words index. The question-and-answer retrieval module 220 can mark part-of-speech and bag-of-words index for the vocabulary in the pre-processed first instant message based on natural language processing. In one embodiment, the question-and-answer retrieval module 220 can mark the part-of-speech for vocabulary according to a pre-stored reference dictionary. In one embodiment, the question-answer retrieval module 220 can mark the vocabulary with a bag-of-words index based on the bag-of-words model. The question-and-answer retrieval module 220 can also analyze the context relationship of each vocabulary in the first instant messaging message based on statistical methods such as word frequency-reverse document frequency, etc., to mark the vocabulary with a bag-of-words index.

In step S604 , the question and answer retrieval module 220 can calculate the similarity between the first instant messaging message and each question message in the question and answer lexicon 260 . If the similarity between the specific question message in the question and answer dictionary 260 and the first instant messaging message (ie: the maximum similarity) is greater than the similarity between other question messages and the first instant messaging message, then the question and answer retrieval module 220 can determine that the specific question message matches the first instant messaging message. Accordingly, the question-answer retrieval module 220 may skip step S605 and directly execute step S606. In step S606, the question-and-answer retrieval module 220 may select 260 to select the answer message corresponding to the specific question message, and reply the second instant messaging message including the selected answer message to the first terminal device 20 through the transceiver 300 .

If there are multiple question messages corresponding to the maximum similarity in the question-answer lexicon 260, the question-answer retrieval module 220 may not be able to determine the selected answer message according to the similarity between the first instant messaging message and each question message. Accordingly, the question and answer retrieval module 220 can execute step S605. In step S605 , the question and answer retrieval module 220 can calculate the similarity between the first instant messaging message and each answer message in the question and answer lexicon 260 . If the similarity between the specific answer message in the question-and-answer lexicon 260 and the first instant messaging message (that is, the maximum similarity) is greater than the similarity between other answer messages and the first instant messaging message, then the question-and-answer retrieval module 220 can determine that the specific answer message matches the first instant messaging message. In step S606, the question and answer retrieval module 220 can select the specific answer message from the question and answer lexicon 260 as the selected answer message, and can reply the second instant messaging message including the selected answer message to the first terminal device 20 through the transceiver 300 .

In one embodiment, the question and answer retrieval module 220 can calculate the similarity between the first instant messaging message and the information in the question and answer lexicon 260 (for example: question message or answer message) based on the best match 25 algorithm (best match 25, BM25). The question-and-answer retrieval module 220 can calculate the similarity according to the part of speech and the bag-of-words index of words in the messages (ie, the first instant messaging message and each message in the question-and-answer thesaurus 260 ). Specifically, in order to calculate the similarity between the first instant messaging message and the specific message in the question and answer lexicon 260, the question and answer retrieval module 220 can perform The algorithm calculates a lexical similarity score, a part-of-speech similarity score, and a bag-of-words similarity score between the vocabulary in the first instant messaging message and the vocabulary in the specific message. The question-answer retrieval module 220 can calculate the similarity according to the vocabulary similarity score, the part-of-speech similarity score and the bag-of-words similarity score. For example, the question-answer retrieval module 220 can calculate the similarity according to a polynomial including vocabulary similarity scores, part-of-speech similarity scores, and bag-of-words similarity scores.

After the automatic answering system 10 replies the second instant messaging message to the customer, the automatic answering system 10 can judge whether to transfer the customer to the customer service personnel according to the customer's emotional response to the second instant messaging message. FIG. 7 shows a flow chart of transferring customer service personnel according to an embodiment of the present invention.

In step S701 , the customer service processing module 240 can obtain a third instant messaging message corresponding to the second instant messaging message from the first terminal device 20 through the transceiver 300 .

In step S702, the customer service processing module 240 can count the count value corresponding to the third instant messaging message. Specifically, the customer service processing module 240 may increase the count value corresponding to the third instant messaging message in response to the third instant messaging message matching the vocabulary in the negative emotion vocabulary 270 . In other words, if there are more words in the negative emotion lexicon 270 in the third instant messaging message, the count value of the third instant messaging message will be larger.

In step S703, the customer service processing module 240 may determine whether the count value is greater than a threshold. If the count value is less than or equal to the threshold value, the customer service processing module 240 may not execute step S704. If the count value is greater than the threshold, then the customer service processing module 240 can determine that the customer service The user is in a negative emotional state. Accordingly, in step S704, the customer service processing module 240 can send a notification message to the second terminal device through the transceiver 300, so as to notify the customer service personnel that the customer is in a negative emotional state. In one embodiment, the customer service processing module 240 can further establish a communication channel between the first terminal device 20 and the second terminal device. In other words, the customer service processing module 240 can automatically transfer the customer to customer service staff when the customer is in a negative emotional state.

FIG. 8 shows a flow chart of an instant messaging service-based automatic answering method according to an embodiment of the present invention, wherein the automatic answering method can be implemented by the automatic answering system 10 shown in FIG. 1 . In step S801, a question-and-answer lexicon and a negative emotion lexicon are obtained, wherein the question-answer lexicon includes a first question message and a first answer message corresponding to the first question message, and the negative emotion lexicon includes a first vocabulary. In step S802, a first instant messaging message is obtained from the first terminal device. In step S803, replying to the first terminal device a second instant messaging message including the first answer message in response to the matching of the first instant messaging message with the first question message. In step S804, a third instant messaging message corresponding to the second instant messaging message is obtained from the first terminal device. In step S805, a notification message is sent to the second terminal device in response to the third instant messaging message matching the first vocabulary.

To sum up, the automatic answering system of the present invention can provide the necessary functions of the question-and-answer robot service, such as unstructured format data cleaning, file key content sorting, question-and-answer sentence normalization, word segmentation/segmentation processing or custom reference dictionary, etc. The automatic answering system can also generate answers with high similarity with the customer's question information through the question and answer retrieval system, so as to achieve the purpose of correctly responding to the question information. Customer service with automated response system The processing module can analyze the customer's emotions, and actively remind the real customer service staff to intervene in the conversation between the user and the question-and-answer robot when the customer's negative emotions exceed the threshold. In addition, in the question-and-answer process, the user can also actively request the system to be served by a real person, so as to achieve a better service experience.

S801, S802, S803, S804, S805: steps

Claims (10)

An automatic answering system based on an instant messaging service, comprising: a transceiver, communicatively connected to a first terminal device and a second terminal device; a storage medium, storing a plurality of modules; and a processor, coupled to the storage medium and the transceiver, and accessing and executing the plurality of modules, wherein the plurality of modules include: a negative emotion lexicon, storing a first vocabulary; a question-and-answer lexicon, storing a first question message, a first answer message corresponding to the first question message, and a second answer message corresponding to the first question message; a message receiving module, through the transceiver. The first terminal device obtains the first instant messaging message; the question and answer retrieval module calculates a first similarity between the first instant messaging message and the first answer message and a second similarity between the first instant messaging message and the second answer message; and responds to the first instant messaging message matching with the first question message and the first similarity is greater than the second similarity, and replies a second instant messaging message including the first answer message to the first terminal device through the transceiver; a third instant messaging message of said second instant messaging message, and A notification message is sent to the second terminal device through the transceiver in response to the third instant messaging message matching the first vocabulary. The automatic answering system according to claim 1, wherein the customer service processing module establishes a communication channel between the first terminal device and the second terminal device in response to the matching of the third instant messaging message with the first vocabulary. The automatic answering system as described in claim 1, wherein the question-answer lexicon further includes a second question message, wherein the question-answer retrieval module calculates a third similarity between the first instant messaging message and the first question message and a fourth similarity between the first instant messaging message and the second question message, wherein the question-answer retrieval module determines that the first instant messaging message matches the first question message in response to the third similarity being greater than the fourth similarity. The automatic answering system according to claim 3, wherein the question and answer retrieval module calculates the third similarity based on a best matching algorithm. The automatic answering system according to claim 3, wherein the question-answer retrieval module marks the second vocabulary in the first question message with a part-of-speech and a bag-of-words index based on natural language processing, wherein the question-answer retrieval module calculates the third similarity according to the part-of-speech and the bag-of-words index. The automatic answering system as described in claim 3, wherein the question-answer retrieval module pre-stores a reference dictionary, and in response to the second vocabulary in the first question message matching a third vocabulary in the reference dictionary, marks the second vocabulary with a part-of-speech and a bag-of-words index according to the third vocabulary, wherein the question-answer retrieval module calculates the third similarity according to the part-of-speech and the bag-of-words index. The automatic answering system as described in claim 1, wherein the plurality of modules further include: a data processing module, which obtains a file containing the first question message and the first answer message through the transceiver, wherein the data processing module generates a mapping relationship between the first question message and the first answer message according to the file, and establishes the question-and-answer lexicon according to the mapping relationship. The automatic answering system as described in claim 1, wherein the customer service processing module obtains the file through the transceiver, and marks the emotional state for the first vocabulary in the file based on natural language processing, wherein the customer service processing module adds the first vocabulary to the negative emotion vocabulary in response to the emotional state being a negative emotional state. The automatic answering system according to claim 1, wherein the customer service processing module increments a count value in response to the third instant messaging message matching the first vocabulary, and sends the notification message in response to the count value being greater than a threshold. An automatic answering method based on an instant messaging service, comprising: obtaining a question-and-answer lexicon and a negative emotion lexicon, wherein the question-answer lexicon includes a first question message, a first answer message corresponding to the first question message, and a second answer message corresponding to the first question message, wherein the negative emotion lexicon includes a first vocabulary; obtaining a first instant messaging message from a first terminal device; two similarities; in response to the first instant messaging message matching the first question message and the first similarity being greater than the second similarity, replying a second instant messaging message including the first answer message to the first terminal device; obtaining a third instant messaging message corresponding to the second instant messaging message from the first terminal device; and sending a notification message to a second terminal device in response to the third instant messaging message matching the first word.

Images