A method and device for intent recognition by filling slots

Abstract

The present invention discloses a method and device for intention recognition through slot filling, which relates to the field of computer technology. A specific implementation manner of the method includes: determining the scenario of the user input information, and then obtaining the standard question forms of each service slot within the scenario; determining the similarity between each standard question form and the user input information, and obtaining the standard question form with the highest similarity; taking the service slot value corresponding to the standard question form whose similarity meets the preset condition as the service slot value of the user input information, so as to recognize the intention of the user input information. This implementation manner can cope with business policy changes through service slot recognition and slot filling, avoid model modification, improve the accuracy and stability of the model, and reduce the operation cost.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular, to a method and apparatus for intent recognition through slot filling. Background Art

[0002] In the process of answering by an unmanned customer service robot in an after-sales scenario, the main answering process is intent recognition (NLU) and solution answering; for example, when a user says "I want to modify my order", the intent recognition result is "modify order", and the solution answering matches the corresponding solution answer "The order has been shipped and is in the distribution process, and the order attributes cannot be modified".

[0003] When business policies change, for example, in the refined answering of the order scenario, the business policies for modifying orders change, and the policies for modifying the delivery address and phone number of the order change. The corresponding answering answers also need to change, and the answering strategies for modifying other attributes of the order need to be distinguished. The existing implementation methods require the model to separately identify sub-intents in the scenario, and it is necessary to separately re-label the corpus of each sub-intent and re-train the model, so as to distinguish the answers based on different classifications.

[0004] In the process of implementing the present invention, the inventors found that there are at least the following problems in the prior art:

[0005] 1. It is necessary to label conversations, train models, etc., and the model training cycle is long;

[0006] 2. The changes in business policies may occur repeatedly. For example, this policy only takes effect in the short term, and it is necessary to go online quickly, go offline quickly after expiration, and restore the original classification model. If the existing method is used, it will bring great challenges and adverse effects to the accuracy and stability of the model;

[0007] 3. Adjusting the model requires a large amount of operation costs. Summary of the Invention

[0008] In view of this, embodiments of the present invention provide a method and apparatus for intent recognition through slot filling, which can respond to business policy changes through business slot recognition and slot filling, avoid model modification, improve the accuracy and stability of the model, and reduce operation costs.

[0009] To achieve the above object, according to one aspect of the embodiments of the present invention, there is provided a method for intent recognition through slot filling, including:

[0010] Determine the scenario of the user input information, and then obtain the standard questions of each business slot within the scenario;

[0011] Determine the similarity between each standard question and the user input information, and obtain the standard question with the highest similarity;

[0012] Use the service slot values corresponding to the standard questions whose similarity meets the preset conditions as the service slot values of the user input information to identify the intention of the user input information.

[0013] Optionally, before the step of obtaining the standard questions of each service slot in the scenario, it further includes:

[0014] Configure the standard questions in the scenario and the service slot values of each standard question.

[0015] Optionally, after the step of configuring the standard questions in the scenario, it further includes:

[0016] Configure the response content corresponding to the standard question; compare the response content with each service slot to determine the service slot value that matches the response content; based on the service slot value that matches the response content, send a prompt message to determine whether to use the service slot value that matches the response content as the service slot value of the standard question and answer.

[0017] Optionally, after the step of configuring the standard questions in the scenario, it further includes:

[0018] Configure the response content corresponding to the standard question; identify the keywords in the response content; obtain the key content corresponding to the keywords from the response content; based on the key content corresponding to the keywords, send a prompt message to determine whether to configure a new service slot value.

[0019] According to the second aspect of the embodiments of the present invention, there is provided a device for intention recognition by filling slots, including:

[0020] A preliminary screening module that determines the scenario of the user input information and then obtains the standard questions of each service slot in the scenario;

[0021] A matching module that determines the similarity between each standard question and the user input information and obtains the standard question with the highest similarity;

[0022] An identification module that uses the service slot values corresponding to the standard questions whose similarity meets the preset conditions as the service slot values of the user input information to identify the intention of the user input information.

[0023] Optionally, the device according to the embodiments of the present invention further includes: a configuration module for: before the step where the preliminary screening module obtains the standard questions of each service slot in the scenario, configuring the standard questions in the scenario and the service slot values of each standard question.

[0024] Optionally, the configuration module is further configured to: after configuring the standard question forms in the scenario, configure the corresponding response content for the standard question forms; compare the response content with each service slot to determine the service slot values that match the response content; and send a prompt message based on the service slot values that match the response content to determine whether to use the service slot values that match the response content as the service slot values of the standard question and answer.

[0025] Optionally, the configuration module is further configured to: after configuring the standard question forms in the scenario, configure the corresponding response content for the standard question forms; identify the keywords in the response content; obtain the key content corresponding to the keywords from the response content; and send a prompt message based on the key content corresponding to the keywords to determine whether to configure new service slot values.

[0026] According to a third aspect of an embodiment of the present invention, there is provided an electronic device for intent recognition by filling slots, including:

[0027] One or more processors;

[0028] A storage device for storing one or more programs,

[0029] When the one or more programs are executed by the one or more processors, the one or more processors implement the method provided in the first aspect of the embodiment of the present invention.

[0030] According to a fourth aspect of an embodiment of the present invention, there is provided a computer-readable medium having a computer program stored thereon, and when the program is executed by a processor, the method provided in the first aspect of the embodiment of the present invention is implemented.

[0031] One of the above embodiments of the invention has the following advantages or beneficial effects: By providing dynamically configurable sub-intent recognition through service slot recognition and slot filling, it can cope with changes in business policies, avoid model modifications, improve the accuracy and stability of the model, and reduce operating costs. When the response content corresponding to the standard question form matches the existing service slot values, a prompt message is sent, which can avoid the problem of high labor costs caused by the need to analyze the response content corresponding to the standard question form one by one, promote the rapid configuration of service slots in new scenarios, and improve operating efficiency. When the response content corresponding to the standard question form contains keywords, a prompt message is sent, which can avoid the problem of missing sub-intent guidance and accurate responses in business scenarios caused by missing service slot value configuration, promote the rapid configuration of service slots in new scenarios, and improve operating efficiency.

[0032] The further effects of the above non-conventional optional methods will be described below in combination with specific embodiments. Description of the Drawings

[0033] The accompanying drawings are used to better understand the present invention and do not constitute an undue limitation on the present invention. Among them:

[0034] Figure 1 is a schematic diagram of the main process of the method for intent recognition by slot filling according to an embodiment of the present invention;

[0035] Figure 2 is a schematic diagram of the process of the method for intent recognition by slot filling in an alternative embodiment of the present invention;

[0036] Figure 3 is a schematic diagram of the main modules of the device for intent recognition by slot filling according to an embodiment of the present invention;

[0037] Figure 4 is an exemplary system architecture diagram to which an embodiment of the present invention can be applied;

[0038] Figure 5 is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention. Detailed Embodiments

[0039] The following describes exemplary embodiments of the present invention with reference to the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for the sake of clarity and conciseness, the description below omits the description of well-known functions and structures.

[0040] The main function of the slot filling logic is to trigger slot filling within a scenario, identify slot values, and finally accurately locate the user's intent and provide an accurate solution. Entity recognition refers to Named Entity Recognition (NER), such as entities like address, color, brand, category, order number, mobile phone number, time, product number, etc. Slot recognition mainly calls NER for recognition. For example, when the user states "I want to modify the order, order number XXXXX", after entering the scenario of modifying the order, the slot filling logic is triggered, calling NER to recognize the order slot, and NER returns the order number and the slot filling is successful, and an accurate response is given based on the order number for the solution response.

[0041] The present invention uses the slot filling logic to implement sub-intent recognition based on configuration: expanding slot recognition, adding a business slot recognition function to slot recognition, and using business slot recognition and slot filling to achieve sub-intent recognition. Business slot recognition refers to the recognition of non-named entities. For example, the scenario of modifying an order is divided into multiple sub-businesses: modifying the order address, modifying the order phone number, modifying the order delivery time, etc.

[0042] According to one aspect of the embodiments of the present invention, a method for intent recognition by slot filling is provided.

[0043] Figure 1 is a schematic diagram of the main process of the method for intent recognition by slot filling according to the embodiments of the present invention, as Figure 1 shown, the method for intent recognition by slot filling includes: step S101, step S102, and step S103.

[0044] Step S101: Determine the scenario of the user input information, and then obtain the standard question forms of each business slot within the scenario.

[0045] The user input information refers to the information input by the user, which can be the user's question information, such as "Can this product be repaired?" and "Where is the origin of this product?"; it can also be other information input by the user, such as "I want to modify the order address" and "I want to cancel the order", etc.

[0046] The intent of the user input information can include two aspects. On the one hand, it is the scenario intent, that is, what type of scenario the user input information belongs to. For example, the scenario type of the user input information "Where is the origin of this product?" is order consultation. On the other hand, it is the sub-intent under the scenario, such as the sub-operations in the order modification scenario: modify order address, modify order phone number, modify order delivery time, etc. For example, the scenario type of the user input information "I want to modify the order address" is order modification, and the sub-intent is to modify the order address.

[0047] In the present invention, the standard question forms of each scenario are pre-saved. First, determine the scenario of the user input information, and then obtain the standard question forms of each business slot within the scenario, which can narrow the matching range in step S102 and improve the matching performance and the performance of the matching model execution in step S102.

[0048] Exemplarily, the standard question forms within the scenario can be searched through ElasticSearch. Elasticsearch is a search engine built on the full-text search engine Apache Lucene(TM). Through the word segmentation ability of Elasticsearch and the ability of search based on word segmentation, a large number of standard question forms are filtered (the feature of ES is high-performance full-text search, supporting fast matching of a large amount of data). Searching for the standard question forms within the scenario through ElasticSearch can further narrow the matching range in step S102 and improve the matching performance and the performance of the matching model execution in step S102. Table 1 shows the scenario of the user input information and the standard question forms filtered within the scenario.

[0049] Table 1 Scenario of user input information and corresponding standard question forms

[0050]

[0051]

[0052] Step S102: Determine the similarity between each of the standard question formulations and the user input information, and obtain the standard question formulation with the highest similarity.

[0053] In this step, compare each of the standard question formulations determined in step S101 with the user input information to calculate the similarity. The similarity is used to measure the degree of similarity between the standard question formulation and the user input information. The similarity algorithm can be selectively determined according to the actual situation, such as a similarity algorithm based on text similarity, a similarity algorithm based on semantic similarity, etc.

[0054] In an alternative embodiment, a similarity matching model (hereinafter referred to as the DSSM model) can be pre-trained. Optionally, the DSSM model is based on semantic similarity matching, manually marks the semantics of the corpus, and performs supervised training based on the manually labeled samples, without the need to perform an unsupervised model mapping in the intermediate process, so the accuracy is relatively high. After training the DSSM model, calculate the distance of the semantic vectors of the user input information and the standard question formulation, so as to achieve the effect of calculating the similarity; if there are multiple standard question formulations, it is necessary to compare the user input information with multiple standard question formulations.

[0055] Step S103: Use the service slot values corresponding to the standard question formulations whose similarity meets the preset conditions as the service slot values of the user input information to identify the intent of the user input information.

[0056] The preset conditions can be selectively determined according to the actual situation. For example, the preset condition is the maximum similarity, or the similarity exceeds a preset similarity threshold. Of course, the preset condition can also be a combination of the two. In the example shown in Table 2, the preset condition is a combination of the above two conditions.

[0057] Table 2 Matching results of user input information

[0058]

[0059] According to the intent identified in step S103, subsequent analysis and processing can be performed. For example, the user's preferences can be analyzed, and when applied to automatic answering, the corresponding answer content for the user input information can also be automatically determined according to the identified intent. Table 3 shows an example of the intent recognition result of the user input information and the corresponding answer content.

[0060] Table 3 Intent recognition result of user input information and corresponding answer content

[0061]

[0062] In the present invention, the user input information is classified by NLU (intent recognition), and enters the scenario according to the classification result. At this time, the slot values of the business slots that need to be recognized in the scenario can be obtained. Then, the user question is matched with multiple standard questions of the slot values of the business slots. First, a full-text search is performed, and the matched standard questions are returned. Then, similarity sorting is performed through an algorithm, and the slot values of the business slots corresponding to the standard questions whose similarity meets the preset conditions are used as the business sub-intents of the user input information. By providing dynamically configurable sub-intent recognition through business slot recognition and slot filling, it can cope with changes in business policies, avoid model modification, improve the accuracy and stability of the model, and reduce operation costs.

[0063] Before the step of obtaining the standard questions of each business slot in the scenario in step S101, it may further include: configuring the standard questions in the scenario and the slot values of the business slots for each standard question. The standard questions in the scenario and the slot values of the business slots for each standard question can be determined according to the actual situation. For example, they can be determined according to the questions of users in the online artificial customer service conversation or the predicted questions by humans, etc. Of course, the slot values of the standard questions of the business slots can also be determined automatically to improve the configuration efficiency, reduce manual consumption and costs.

[0064] In some embodiments, the slot values of the business slots for each standard question are determined according to the following steps: obtaining the response content corresponding to the standard question; comparing the response content with the existing business slots to determine the slot value of the business slot that matches the response content (for example, having the maximum similarity or the similarity being greater than the preset threshold); using the slot value of the business slot that matches the response content as the slot value of the current standard question. By filling the slot in this way, it can avoid the problem of high labor costs caused by the need to analyze the response content corresponding to each standard question one by one, promote the rapid configuration of business slots in new scenarios, and improve operation efficiency.

[0065] In other embodiments, the business slot value of each standard question is determined according to the following steps: obtaining the answer content corresponding to the standard question; identifying the keywords in the answer content; obtaining the key content corresponding to the keyword from the answer content; issuing a reminder based on the key content to prompt the operator whether a new business slot value needs to be created for the current standard question. Keywords can be defined according to actual conditions, for example, the answer content of each standard question that has been sorted out is analyzed, and then the leading words in the universal answer content are sorted out to form keywords, such as "if", "if", "you can", "whether", etc. The key content can refer to a sentence containing keywords, such as the content corresponding to the leading word. If these conditional leading words exist, it means that there is a potential business slot, and the content corresponding to the keyword is displayed to the operator, and the operator decides whether to configure a new business slot. Filling the slot in this way can avoid the problem of missing the sub-intention guidance and accurate response of the business scenario due to missing the business slot value configuration, promote the rapid configuration of the business slot in the new scenario, and improve the operation efficiency.

[0066] Optionally, after the step of configuring the standard questions in the scenario in step S102, it also includes: configuring the answer content corresponding to the standard questions; comparing the answer content with each business slot to determine the business slot value that matches the answer content; issuing a prompt message based on the business slot value that matches the answer content to determine whether to use the business slot value that matches the answer content as the business slot value of the standard question. By issuing a prompt message, the business slot value of the configured standard question can be automatically identified, avoiding the problem of high labor costs caused by the need to analyze the answer content corresponding to the standard questions one by one, promoting the rapid configuration of business slots in new scenarios, and improving operational efficiency. Exemplarily, the answer content and prompt information for configuring standard questions are shown in Table 4.

[0067] Table 4: Response content and prompt information for configuring standard questions

[0068]

[0069] Optionally, after the step of configuring the standard question in step S102, the following steps are further included: configuring the response content corresponding to the standard question; identifying the keywords in the response content; obtaining the key content corresponding to the keyword from the response content; and sending a prompt message based on the key content corresponding to the keyword to determine whether to configure a new business slot value. Generally, the verification of the response content effect by the operator needs to go through processes such as first going live with the response content, then looking at user feedback, then optimizing the response content, manually sorting out the response content and discovering business slots. The cycle of this process is generally long, and the existing response content may not be sorted out during the sorting. Therefore, recommending business slots by means of automatically sending prompt messages is an important part of improving efficiency, which can avoid problems such as missing sub-intent guidance and accurate response of business scenarios caused by missing business slot value configuration, promote the rapid configuration of business slots in new scenarios, and improve operation efficiency. Exemplarily, the response content and prompt messages for configuring the standard question are shown in Table 5.

[0070] Table 5 Response Content and Prompt Messages for Configuring the Standard Question

[0071]

[0072] The present invention provides dynamically configurable sub-intent recognition through the methods of business slot recognition and slot filling, which can cope with changes in business policies, avoid model modification, improve the accuracy and stability of the model, and reduce operation costs. Sending a prompt message when the response content corresponding to the standard question matches the existing business slot value can avoid the problem of high labor costs caused by the need to analyze the response content corresponding to the standard question one by one, promote the rapid configuration of business slots in new scenarios, and improve operation efficiency. Sending a prompt message when the response content corresponding to the standard question contains keywords can avoid the problems of missing sub-intent guidance and accurate response of business scenarios caused by missing business slot value configuration, promote the rapid configuration of business slots in new scenarios, and improve operation efficiency.

[0073] The following Figure 2 is an exemplary illustration of the method of the present invention with application to scenario response as an example. Figure 2 is a schematic flowchart of the method for realizing intent recognition through slot filling in an optional embodiment of the present invention. As Figure 2 shown, the method for realizing intent recognition through slot filling mainly includes two aspects:

[0074] I. Business Slot Configuration

[0075] There are two ways to configure business slots:

[0076] (1) After the operator inputs the answer (i.e., the response content of the standard question), the service recognition module (i.e., the module that executes steps S101, S102, and S103 of the present invention) is called, and the service slots are matched using the keywords in the answer. If a match is found, it indicates that there are service slots, and a prompt message is sent to remind the operator that there are service slots. The operator configures new service slots based on the reminder information.

[0077] (2) The operator directly configures the service slots.

[0078] II. Service Slot Recognition

[0079] After receiving the user input information and entering the scenario, the steps of scenario service slot recognition are entered. In this step, the service slot recognition module is called to perform the following process:

[0080] The standard questions are filtered through ElasticSearch to obtain the standard questions corresponding to the user input information; if the filtering result is not empty, it indicates that there are results. The filtered results are sorted through DSSM, and it is determined whether there are results higher than the threshold according to the preset conditions. If there are, the service slot values corresponding to the results are obtained and returned to achieve slot filling. The scenario response is performed according to the returned service slot values and the response content is output.

[0081] In this example, ElasticSearch and DSSM models are used to recognize service slots. Through the service slot recognition method, dynamically configurable sub-intent recognition is provided, and at the same time, automatic prompting of potential service slots in the response content is provided, ensuring the stability of the model, reducing the maintenance cost, and improving the operation efficiency.

[0082] According to the second aspect of the embodiments of the present invention, there is provided an apparatus for implementing the method provided in the first aspect of the embodiments of the present invention.

[0083] Figure 3 It is a schematic diagram of the main modules of the apparatus for intent recognition through slot filling according to the embodiments of the present invention. As Figure 3 shown, the apparatus 300 for intent recognition through slot filling includes:

[0084] The preliminary screening module 301 determines the scenario of the user input information, and then obtains the standard questions of each service slot within the scenario;

[0085] The matching module 302 determines the similarity between each of the standard questions and the user input information, and obtains the standard question with the highest similarity;

[0086] The recognition module 303 uses the service slot values corresponding to the standard questions whose similarity meets the preset conditions as the service slot values of the user input information to recognize the intent of the user input information.

[0087] Optionally, the device according to an embodiment of the present invention further includes: a configuration module, configured to: before the step of the preliminary screening module obtaining the standard question forms of each service slot in the scenario, configure the standard question forms in the scenario, and the service slot values of each standard question form.

[0088] Optionally, the configuration module is further configured to: after the step of configuring the standard question forms in the scenario, configure the corresponding response content for the standard question forms; compare the response content with each service slot to determine the service slot values that match the response content; and send a prompt message based on the service slot values that match the response content to determine whether to use the service slot values that match the response content as the service slot values of the standard question and answer.

[0089] Optionally, the configuration module is further configured to: after the step of configuring the standard question forms in the scenario, configure the corresponding response content for the standard question forms; identify the keywords in the response content; obtain the key content corresponding to the keywords from the response content; and send a prompt message based on the key content corresponding to the keywords to determine whether to configure new service slot values.

[0090] According to a third aspect of an embodiment of the present invention, there is provided an electronic device for realizing intent recognition by slot filling, including:

[0091] One or more processors;

[0092] A storage device for storing one or more programs,

[0093] When the one or more programs are executed by the one or more processors, the one or more processors implement the method provided in the first aspect of the embodiment of the present invention.

[0094] According to a fourth aspect of an embodiment of the present invention, there is provided a computer-readable medium, on which a computer program is stored, and when the program is executed by a processor, the method provided in the first aspect of the embodiment of the present invention is implemented.

[0095] Figure 4 An exemplary system architecture 400 to which the method for realizing intent recognition by slot filling or the device for realizing intent recognition by slot filling according to the embodiment of the present invention can be applied is shown.

[0096] As Figure 4 shown, the system architecture 400 may include terminal devices 401, 402, 403, a network 404, and a server 405. The network 404 is used to provide a medium for a communication link between the terminal devices 401, 402, 403 and the server 405. The network 404 may include various connection types, such as wired, wireless communication links, or fiber optic cables, etc.

[0097] Users can use terminal devices 401, 402, and 403 to interact with server 405 via network 404 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 401, 402, and 403, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (for example only).

[0098] Terminal devices 401, 402, and 403 can be various electronic devices with a display screen and supporting web browsing, including but not limited to smartphones, tablets, laptop computers, desktop computers, and so on.

[0099] Server 405 can be a server that provides various services, such as a background management server that supports shopping websites browsed by users using terminal devices 401, 402, and 403 (for example only). The background management server can analyze and process data such as received question-and-answer requests, etc., and feedback the processing results (such as the response content determined based on the recognized intent - for example only) to the terminal device.

[0100] It should be noted that the method for realizing intent recognition through slot filling provided by the embodiments of the present invention is generally executed by server 405. Correspondingly, the device for realizing intent recognition through slot filling is generally set in server 405.

[0101] It should be understood that Figure 4 the numbers of terminal devices, networks, and servers in

[0102] are merely illustrative. According to the implementation requirements, there can be any number of terminal devices, networks, and servers. Figure 5 Shown below is a schematic structural diagram of a computer system 500 of a terminal device suitable for implementing the embodiments of the present invention with reference to Figure 5 The shown terminal device is merely an example and should not impose any limitations on the functions and usage scope of the embodiments of the present invention.

[0103] As Figure 5 shown, computer system 500 includes a central processing unit (CPU) 501, which can execute various appropriate actions and processes according to the program stored in the read-only memory (ROM) 502 or the program loaded from the storage section 508 into the random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of system 500 are also stored. The CPU 501, ROM 502, and RAM 503 are connected to each other via a bus 504. The input / output (I / O) interface 505 is also connected to the bus 504.

[0104] The following components are connected to the I / O interface 505: an input section 506 including a keyboard, a mouse, etc.; an output section 507 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc. and a speaker, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, a modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the I / O interface 505 as needed. A removable medium 511 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc. is mounted on the drive 510 as needed so that a computer program read therefrom is installed into the storage section 508 as needed.

[0105] Specifically, according to the embodiments disclosed by the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed by the present invention include a computer program product which includes a computer program carried on a computer-readable medium, and the computer program includes program codes for performing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 509, and / or installed from the removable medium 511. When the computer program is executed by a central processing unit (CPU) 501, the above-described functions defined in the system of the present invention are executed.

[0106] It should be noted that the computer-readable medium shown in the present invention can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of a computer-readable storage medium can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present invention, a computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on a computer-readable medium can be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination of the above.

[0107] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks can occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown can actually be executed substantially in parallel, and they can sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram or flowchart, and the combination of blocks in the block diagram or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0108] The modules involved in the embodiments of the present invention can be implemented in software or in hardware. The described modules can also be provided in a processor. For example, it can be described as: A processor includes: a primary screening module that determines the scenario of the user input information and then obtains the standard questions of each service slot within the scenario; a matching module that determines the similarity between each of the standard questions and the user input information and obtains the standard question with the highest similarity; and an identification module that uses the service slot value corresponding to the standard question whose similarity meets a preset condition as the service slot value of the user input information to identify the intent of the user input information. Among them, the names of these modules do not constitute a limitation on the module itself in some cases. For example, the primary screening module can also be described as "the module for identifying the intent of the user input information".

[0109] On the other hand, the present invention also provides a computer-readable medium, which can be included in the device described in the above embodiments; or it can exist alone without being assembled into the device. The above computer-readable medium carries one or more programs. When the above one or more programs are executed by a device, the device includes: determining the scenario of the user input information and then obtaining the standard questions of each service slot within the scenario; determining the similarity between each of the standard questions and the user input information and obtaining the standard question with the highest similarity; and using the service slot value corresponding to the standard question whose similarity meets a preset condition as the service slot value of the user input information to identify the intent of the user input information.

[0110] According to the technical solution of the embodiments of the present invention, by means of service slot identification and slot filling, it provides dynamically configurable sub-intent identification, can cope with business policy changes, avoid model modification, improve the accuracy and stability of the model, and reduce operation costs. When the response content corresponding to the standard question matches the existing service slot value, a prompt message is sent, which can avoid the problem of high labor costs caused by the need to analyze the response content corresponding to each standard question one by one, and promote the rapid configuration of service slots in new scenarios, improving operation efficiency. When the response content corresponding to the standard question contains keywords, a prompt message is sent, which can avoid the problem of missing sub-intent guidance and accurate response of business scenarios caused by missing service slot value configuration, promote the rapid configuration of service slots in new scenarios, and improve operation efficiency.

[0111] The above specific implementation manners do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for intent recognition by filling slots, characterized in that, Including: Determine the scenario intention of the user input information, and then obtain the standard questions of each business slot in the scenario corresponding to the scenario intention; Determine the similarity between each of the standard questions and the user input information, and obtain the standard question with the highest similarity; Use the business slot values corresponding to the standard questions whose similarity meets the preset conditions as the business sub-intentions of the user input information in the scenario to identify the intention of the user input information; The intention of the user input information includes: the scenario intention and the business sub-intentions under the scenario intention; Before obtaining the standard questions of each business slot in the scenario corresponding to the scenario intention, it also includes: configuring the standard questions in the scenario, configuring the response content corresponding to the standard questions; comparing the response content with each business slot to determine the business slot value that matches the response content, and sending a prompt message based on the business slot value that matches the response content to determine whether to use the business slot value that matches the response content as the business slot value of the standard question and answer; or, identifying the keywords in the response content; obtaining the key content corresponding to the keywords from the response content; sending a prompt message based on the key content corresponding to the keywords to determine whether to configure a new business slot value; the prompt message includes: a reminder message that there is a potential business slot in the response content and the corresponding configurable solution branch.

2. The method according to claim 1, characterized in that, Before the step of obtaining the standard questions of each business slot in the scenario corresponding to the scenario intention, it also includes: Configure the business slot values of each of the standard questions.

3. An apparatus for intent recognition by filling slots, characterized in that, Including: A preliminary screening module that determines the scenario intention of the user input information and then obtains the standard questions of each business slot in the scenario corresponding to the scenario intention; A matching module that determines the similarity between each of the standard questions and the user input information and obtains the standard question with the highest similarity; An identification module that uses the business slot values corresponding to the standard questions whose similarity meets the preset conditions as the business sub-intentions of the user input information in the scenario to identify the intention of the user input information; The intention of the user input information includes: the scenario intention and the business sub-intentions under the scenario intention; The configuration module is also used for: before obtaining the standard questions of each business slot in the scenario corresponding to the scenario intention, configuring the standard questions in the scenario, configuring the response content corresponding to the standard questions; comparing the response content with each business slot to determine the business slot value that matches the response content, and sending a prompt message based on the business slot value that matches the response content to determine whether to use the business slot value that matches the response content as the business slot value of the standard question and answer; or, identifying the keywords in the response content; obtaining the key content corresponding to the keywords from the response content; sending a prompt message based on the key content corresponding to the keywords to determine whether to configure a new business slot value; the prompt message includes: a reminder message that there is a potential business slot in the response content and the corresponding configurable solution branch.

4. The device according to claim 3, wherein Also including: Configuration module, for: configuring the slot values of each of the standard questions before the step in which the preliminary screening module obtains the standard questions of each service slot in the scenario.

5. An electronic device for intent recognition by filling slots, characterized in that Comprising: One or more processors; A storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1-2.

6. A computer-readable medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, the method according to any one of claims 1-2 is implemented.

Images