Knowledge answer search method and device

Abstract

The present invention provides a method and apparatus for searching knowledge answers. The method includes: when detecting that a customer calls in, obtaining the customer's telephone number and the calling area; performing a preliminary screening on the knowledge forest based on the calling area to obtain a screened knowledge forest; searching for the historical service record of the telephone number based on the telephone number, and extracting each historical search tag in the historical service record; obtaining the voice information of the customer and converting the voice information into text information; matching the text information with each historical search tag; if the text information matches any historical search tag, using the answer information corresponding to the historical search tag as the answer information corresponding to the text information; otherwise, extracting each question tag in the text information and searching for the answer information corresponding to each question tag in the screened knowledge forest. By applying the method provided by the present invention, answers to multiple questions can be quickly found through voice question and answer and fed back to the customer, improving the customer experience.

Description

Technical Field

[0001] The present invention relates to the technical field of speech recognition, and particularly to a method and device for searching knowledge answers. Background Art

[0002] With the rapid development of Internet technology, it is more convenient and fast for customers to handle various procedures through the Internet. With the online and fast processing of many procedures, when customers handle various procedures and services online, they are usually not familiar with the relevant business processes. Therefore, in order to solve the problem that users are not familiar with the business processes, online application software basically adds a knowledge Q&A service module.

[0003] The knowledge Q&A service modules of the prior art are basically pre-set questions, and customers can view the relevant answers by clicking on the pre-set questions. However, in the prior art, searching for answers has limitations. Customers cannot ask multiple questions at once, and when customers have multiple contents to understand, they need to search for the questions of each content one by one and click to view them, resulting in a poor customer experience. Summary of the Invention

[0004] In view of this, the present invention provides a method for searching knowledge answers. Through this method, the answers to multiple questions can be quickly found through voice Q&A and fed back to customers, improving the customer experience.

[0005] The present invention also provides a device for searching knowledge answers to ensure the implementation and application of the above method in practice.

[0006] A method for searching knowledge answers includes:

[0007] When it is detected that a customer calls in, obtain the customer's telephone number and the calling area;

[0008] Based on the calling area, perform a preliminary screening on a pre-set knowledge forest to obtain a screened knowledge forest;

[0009] Based on the telephone number, search for the historical service records of the telephone number, and extract each historical search tag in the historical service records;

[0010] Obtain the voice information of the customer, and convert the voice information into text information;

[0011] Match the text information with each of the historical search tags;

[0012] If the text information matches any of the historical search tags, use the answer information corresponding to the historical search tag as the answer information corresponding to the text information;

[0013] If the text information does not match all the historical search tags, extract each problem tag in the text information, and according to each problem tag, search for the answer information corresponding to each problem tag in the filtered knowledge forest.

[0014] For the above method, optionally, the initial screening of the pre-set knowledge forest based on the call area to obtain the filtered knowledge forest includes:

[0015] Obtain each sub-tree set in the knowledge forest, each sub-tree corresponds to a different business type, each sub-tree contains multiple branches, and each branch corresponds to a business area;

[0016] Determine each target business type supported by the call area;

[0017] Retain the sub-trees in the knowledge forest corresponding to each target business type, and based on the call area and the business areas corresponding to each branch in each sub-tree, prune the branches of each sub-tree to obtain the filtered knowledge forest.

[0018] For the above method, optionally, the conversion of the voice information into text information includes:

[0019] Remove the noise data in the voice information to obtain denoised voice information;

[0020] Convert the denoised voice information into text content, and identify the context meaning of the text content;

[0021] Based on the context meaning of the text content, correct the text content to obtain the text information corresponding to the voice information.

[0022] For the above method, optionally, the extraction of each problem tag in the text information and the search for the answer information corresponding to each problem tag in the filtered knowledge forest according to each problem tag include:

[0023] Based on the context meaning of the text information, segment the context of the text information;

[0024] Determine the business type to which each segment belongs, and extract each keyword in each segment to obtain a keyword group for each segment;

[0025] Set each keyword group as a problem tag;

[0026] In the filtered knowledge forest, according to the business type corresponding to each problem tag, search for whether there is answer information corresponding to each problem tag;

[0027] If there is answer information corresponding to each of the problem tags, then feedback each of the answer information to the customer;

[0028] If there is no answer information corresponding to any of the problem tags, then prompt the customer to transfer to the artificial seat service.

[0029] Optionally, the above method further includes:

[0030] Based on the context meaning of the text information, delete the stop words in the text information and perform synonym replacement on the synonyms in the text information.

[0031] A knowledge answer search device includes:

[0032] An acquisition unit, configured to acquire the customer's telephone number and the call area when detecting that a customer calls in;

[0033] A screening unit, configured to perform a preliminary screening on a pre-set knowledge forest based on the call area to obtain a screened knowledge forest;

[0034] A first search unit, configured to search for the historical service records of the telephone number based on the telephone number and extract each historical search tag in the historical service records;

[0035] A conversion unit, configured to acquire the customer's voice information and convert the voice information into text information;

[0036] A matching unit, configured to match the text information with each of the historical search tags;

[0037] A second search unit, configured to, if the text information matches any of the historical search tags, use the answer information corresponding to the historical search tag as the answer information corresponding to the text information;

[0038] A third search unit, configured to, if the text information does not match all of the historical search tags, extract each problem tag in the text information and, according to each of the problem tags, search in the screened knowledge forest for the answer information corresponding to each problem tag.

[0039] Optionally, for the above device, the screening unit includes:

[0040] An acquisition subunit, configured to acquire each subtree set in the knowledge forest, each of the subtrees corresponding to a different service type, each subtree including a plurality of branches, and each branch corresponding to a service area;

[0041] A determination subunit, configured to determine each target service type supported by the call area;

[0042] A screening subunit, configured to retain subtrees corresponding to each of the target service types in the knowledge forest, and prune the branches of each subtree based on the calling area and the service areas corresponding to the respective branches in each subtree, to obtain a screened knowledge forest.

[0043] For the above-mentioned device, optionally, the conversion unit includes:

[0044] A denoising subunit, configured to remove noise data in the voice information to obtain denoised voice information;

[0045] A conversion subunit, configured to convert the denoised voice information into text content and recognize the contextual meanings of the text content;

[0046] A correction subunit, configured to correct the text content based on the contextual meanings of the text content to obtain the text information corresponding to the voice information.

[0047] For the above-mentioned device, optionally, the third search unit includes:

[0048] A segmentation subunit, configured to segment the context of the text information based on the contextual meanings of the text information;

[0049] An extraction subunit, configured to determine the service type to which each segment belongs and extract each keyword in each segment to obtain a keyword group for each segment;

[0050] A setting subunit, configured to set each keyword group as a question tag;

[0051] A search subunit, configured to search in the screened knowledge forest for the answer information corresponding to each question tag according to the service type corresponding to each question tag;

[0052] A feedback subunit, configured to, if there is answer information corresponding to each question tag, feedback each answer information to the customer;

[0053] A prompt subunit, configured to, if there is no answer information corresponding to any question tag, prompt the customer to transfer to the artificial seat service.

[0054] For the above-mentioned device, optionally, it further includes:

[0055] A text processing subunit, configured to delete stop words in the text information and perform synonym replacement on synonyms in the text information based on the contextual meanings of the text information.

[0056] A storage medium, the storage medium including stored instructions, wherein when the instructions run, the device where the storage medium is located is controlled to execute the above-mentioned knowledge answer search method.

[0057] An electronic device, including a memory, and one or more instructions, wherein the one or more instructions are stored in the memory and are configured to be executed by one or more processors to execute the above-mentioned knowledge answer search method.

[0058] Compared with the prior art, the present invention has the following advantages:

[0059] The present invention provides a knowledge answer search method, including: when detecting that a customer calls in, obtaining the customer's telephone number and the calling area; based on the calling area, performing a preliminary screening on a pre-set knowledge forest to obtain a screened knowledge forest; based on the telephone number, searching for the historical service record of the telephone number and extracting each historical search tag in the historical service record; obtaining the customer's voice information and converting the voice information into text information; matching the text information with each of the historical search tags; if the text information matches any of the historical search tags, using the answer information corresponding to the historical search tag as the answer information corresponding to the text information; if the text information does not match all of the historical search tags, extracting each question tag in the text information and, according to each question tag, searching in the screened knowledge forest for the answer information corresponding to each question tag. By applying the method provided by the present invention, answers to multiple questions can be quickly found through voice question and answer and fed back to the customer, improving the customer experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0061] Figure 1 It is a method flowchart of a knowledge answer search method provided by an embodiment of the present invention;

[0062] Figure 2 It is another method flowchart of a knowledge answer search method provided by an embodiment of the present invention;

[0063] Figure 3 It is yet another method flowchart of a knowledge answer search method provided by an embodiment of the present invention;

[0064] Figure 4The device structure diagram of a knowledge answer search device provided by an embodiment of the present invention;

[0065] Figure 5 The schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Detailed implementation manners

[0066] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0067] In the present application, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or sequence between these entities or operations. The terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the phrase "comprising a..." does not exclude the existence of additional identical elements in the process, method, article or device including the element.

[0068] The present invention can be used in numerous general-purpose or special-purpose computing device environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, multi-processor devices, distributed computing environments including any of the above devices or equipment, and so on.

[0069] An embodiment of the present invention provides a knowledge answer search method. This method can be applied to a variety of system platforms, and its execution subject can be a processor of a computer terminal or various mobile devices. The method flow chart of the method is as Figure 1 shown, and specifically includes:

[0070] S101: When it is detected that a customer calls in, obtain the phone number and the call area of the customer.

[0071] In the method provided by the embodiment of the present invention, when a customer needs to ask a question about a certain business process, the customer can ask questions by dialing through an application software or directly calling the customer service landline. When it is detected that a customer calls in, obtain the phone number and the call area of the customer.

[0072] Among them, the call area can be the place of origin of the telephone number, or by checking the incoming line, it is determined that the current location of the user is the call area.

[0073] S102: Based on the call area, perform a preliminary screening on the pre-set knowledge forest to obtain the screened knowledge forest.

[0074] In the method provided by the embodiment of the present invention, the knowledge forest contains multiple subtrees, and each subtree corresponds to a business type, such as: services like deposit, loan, credit card, ETC, and corporate remittance. Each subtree contains multiple branches, and each branch corresponds to a business area, such as business areas like Beijing, Shanghai, and Guangzhou. Each branch contains multiple leaves, and each leaf corresponds to a knowledge question and answer information.

[0075] S103: Based on the telephone number, search for the historical service records of the telephone number and extract each historical search tag in the historical service records.

[0076] In the method provided by the embodiment of the present invention, in the actual application process, for the same customer, the questions they have historically consulted may be relatively similar. Therefore, according to the user's telephone number, search for each historical search tag to determine the questions the customer has historically consulted.

[0077] For example, if the customer often consults questions about loans in the historical service records, then when receiving the customer's call subsequently, the customer may still be consulting about loan questions.

[0078] It should be noted that the historical search tag is a keyword group of each keyword of the questions consulted by the customer in the historical service records.

[0079] S104: Obtain the voice information of the customer and convert the voice information into text information.

[0080] In the method provided by the embodiment of the present invention, the customer makes a voice query by making a call.

[0081] Among them, after connecting to the customer's call, the customer is prompted to send voice information in the form of a voice prompt. For example, set a voice prompt tone, and the customer starts voice inputting questions after receiving the prompt tone.

[0082] S105: Match the text information with each of the historical search tags.

[0083] In the method provided by the embodiment of the present invention, the text information is matched with each historical search tag to determine whether the question the customer is consulting this time has been consulted in the historical service records.

[0084] S106: If the text information matches any of the historical search tags, the answer information corresponding to the historical search tag is used as the answer information corresponding to the text information.

[0085] In the method provided by the embodiment of the present invention, if the text information successfully matches at least one historical search tag, the answer information corresponding to the historical search tag is directly used as the current answer information of the customer.

[0086] Among them, for this text information, there may be multiple questions to be consulted. Therefore, the searched answer information is retained first, and after the answer information of other questions is found later, it is sent to the customer uniformly.

[0087] S107: If the text information does not match all of the historical search tags, each question tag in the text information is extracted, and according to each question tag, the answer information corresponding to each question tag is searched in the filtered knowledge forest.

[0088] In the method provided by the embodiment of the present invention, if the text information does not match all the historical search tags, each question tag needs to be extracted, and each question tag is each question asked by the customer. Since the knowledge forest has been preliminarily filtered according to the incoming call area, the number of branches and leaves in each branch of the knowledge forest tree is relatively small. When searching for the answer information corresponding to each question tag, the answer information corresponding to each question tag can be obtained quickly.

[0089] It should be noted that the question tag is a keyword group of each keyword in the question consulted by the customer.

[0090] In the knowledge answer search method provided by the embodiment of the present invention, when it is detected that a customer makes an incoming call, the customer's telephone number and the incoming call area are obtained; and the knowledge forest is preliminarily screened based on the incoming call area, specifically, the subtrees and the branches on the subtrees in the knowledge forest are reduced. Based on the telephone number, each historical search tag of the customer in the historical service record is obtained. For example, the historical search tags of all questions consulted by the user within one month are obtained, and the specific time period involved in obtaining the historical search tags will not be limited here. After the customer dials in, the customer is prompted to input voice information. When the customer's voice information is obtained, the voice information is converted into text information, and the text information is matched with each historical search tag. If the match is successful, the answer information corresponding to the historical search tag is used as the answer information of the text information. If the match fails, the answer information corresponding to each question tag in the text information is searched in the filtered knowledge forest.

[0091] Optionally, after the answer information corresponding to the question tag is found through the knowledge forest, the question tag and its corresponding answer information are saved.

[0092] By applying the method provided in the embodiment of the present invention, the answers to multiple questions can be quickly found through voice Q&A and fed back to the customer, improving the customer experience.

[0093] Based on the content of S102 above, in the method provided by the embodiment of the present invention, the process of initially screening the pre-set knowledge forest based on the calling area to obtain the screened knowledge forest is as Figure 2 shown, and specifically may include:

[0094] S201: Obtain each sub-tree set in the knowledge forest.

[0095] Among them, each of the sub-trees corresponds to a different business type, each sub-tree contains multiple branches, and each branch corresponds to a business area.

[0096] S202: Determine each target business type supported by the calling area.

[0097] In the method provided by the embodiment of the present invention, for each different region, the supported business types are different. For new business types, they are basically first tried out in big city regions such as Beijing, Shanghai, and Guangdong, while other regions may not have launched the new business; or the enterprise adds region-related business according to the characteristics of each region. Therefore, the supported business types in each region can be different.

[0098] S203: Retain the sub-trees in the knowledge forest corresponding to each target business type, and based on the calling area and the business areas corresponding to each branch in each sub-tree, prune the branches of each sub-tree to obtain the screened knowledge forest.

[0099] In the method provided by the embodiment of the present invention, for the same business type, there may be multiple regions that support this business type. Therefore, after retaining the sub-trees corresponding to each target business type supported by the calling area in the knowledge forest, pruning is performed on each sub-tree to obtain the screened knowledge forest.

[0100] In the knowledge answer search method provided by the embodiment of the present invention, during the initial screening of the knowledge forest, each sub-tree set in the knowledge forest is obtained, and each target business type supported by the calling area is determined. In the knowledge forest, the sub-trees corresponding to each target business type are saved, and the branches of each retained sub-tree are pruned according to the calling area to obtain the screened knowledge forest. By screening the knowledge forest, the search progress can be accelerated when answer information is needed in the knowledge forest.

[0101] In the method provided by the embodiments of the present invention, based on the content of S104 above, the process of converting the voice information into text information is as Figure 3 shown, and specifically may include:

[0102] S301: Remove the noise data in the voice information to obtain denoised voice information.

[0103] In the method provided by the embodiments of the present invention, when a customer inputs voice information, it is impossible to ensure that the surrounding environment is quiet. To avoid conversion errors during the process of converting voice information into text information, it is necessary to remove the noise data in the voice information to obtain denoised voice information.

[0104] Among them, a sound spectrum is obtained according to the voice information, and voice recognition software is used to identify the noise data in the sound spectrum to remove the noise data.

[0105] S302: Convert the denoised voice information into text content, and identify the context meaning of the text content.

[0106] In the method provided by the embodiments of the present invention, the denoised voice information is converted into text content, the text content is segmented, and the context meaning of the entire text content is identified.

[0107] It can be understood that when converting voice information into text content, there may be some incorrect conversions of words. For example, "loan" is converted into "bring money". Therefore, it is necessary to identify the context meaning of the text content to determine whether the converted text content is correct.

[0108] S303: Based on the context meaning of the text content, correct the text content to obtain the text information corresponding to the voice information.

[0109] In the embodiments of the present invention, if the converted text content is not correct, then based on the context meaning, correct the text content to obtain the final text information.

[0110] In the knowledge answer search method provided by the embodiments of the present invention, after obtaining the voice information, according to the sound spectrum of the voice information, identify the noise data in the voice information, and remove the noise data to obtain denoised voice information. Convert the denoised voice information into text content, identify the context meaning of the text content, and based on the context meaning, determine whether there are incorrect conversion words. If there are incorrect conversion words, then correct the text content according to the context meaning to obtain the final text information.

[0111] In the method provided by the embodiment of the present invention, based on the content of S107 above, extracting each question tag in the text information, and searching for the answer information corresponding to each question tag in the filtered knowledge forest may specifically include:

[0112] Segment the context of the text information based on the context semantics of the text information;

[0113] Determine the business type to which each segment belongs, and extract each keyword in each segment to obtain a keyword group for each segment;

[0114] Set each keyword group as a question tag;

[0115] In the filtered knowledge forest, search for the answer information corresponding to each question tag according to the business type corresponding to each question tag;

[0116] If there is answer information corresponding to each question tag, feedback each answer information to the customer;

[0117] If there is no answer information corresponding to any question tag, prompt the customer to transfer to the artificial seat service.

[0118] It can be understood that when a customer inputs voice information, the customer may input multiple questions to be consulted. Therefore, after converting the voice information into text information, the text information is segmented, and each segment corresponds to a question consulted by the customer. When the customer consults a question, the customer does not need to be limited to consulting questions of the same business type. Therefore, in the entire voice information, the customer can ask questions of different business types. Extract each keyword in each segment, determine the keyword group corresponding to each segment as a question tag, search for the answer information corresponding to each question tag in the filtered knowledge forest. If the answer information can be found, the answer information is fed back to the customer. Otherwise, the customer is prompted to transfer to the artificial seat.

[0119] In an optional embodiment of the present invention, if the answer information corresponding to the question tag is not found, the customer can be prompted to re-enter the voice information corresponding to the question tag again, and after converting the re-entered voice information into text information, search for the answer information again. If the answer information is still not found after multiple re-entries, then prompt the customer to transfer to the artificial seat service.

[0120] Further, after transferring to the artificial seat service and the artificial provides the answer information to the customer, save the question tag and its corresponding answer information to the subtree corresponding to the business type of the question tag.

[0121] For the method provided by the present invention, the customer only needs to call and input the question to be consulted by voice to obtain the corresponding answer information, which improves the customer experience.

[0122] Optionally, after obtaining the text information, in order to facilitate the extraction of each question tag in the text information, it is necessary to process the text information. The specific processing process includes:

[0123] Based on the context meaning of the text information, stop words in the text information are deleted, and synonyms in the text information are replaced with synonyms.

[0124] Specifically, stop word deletion means removing stop words in the candidate keywords, such as modal particles like "ah", "ma", "ba", "ai", etc. or other words without actual meaning. In actual applications, there are multiple naming methods for the same thing. Therefore, when the customer inputs voice information, they may change the name of a certain thing to other names. To avoid being unable to find the corresponding information of the thing when searching for answer information, it is necessary to replace the synonyms of the thing with the set standard naming method. Therefore, deleting stop words in the text information and replacing synonyms can effectively improve the efficiency of searching for answer information.

[0125] The specific implementation processes and their derivative methods of the above various embodiments are all within the protection scope of the present invention.

[0126] Corresponding to Figure 1 the method described above, an embodiment of the present invention further provides a knowledge answer search device for Figure 1 the specific implementation of the method. The knowledge answer search device provided by the embodiment of the present invention can be applied to a computer terminal or various mobile devices. Its structural schematic diagram is as Figure 4 shown, and specifically includes:

[0127] An acquisition unit 401, configured to acquire the customer's phone number and call area when detecting that the customer calls in;

[0128] A screening unit 402, configured to perform preliminary screening on a pre-set knowledge forest based on the call area to obtain a screened knowledge forest;

[0129] A first search unit 403, configured to search for the historical service record of the phone number based on the phone number and extract each historical search tag in the historical service record;

[0130] A conversion unit 404, configured to acquire the customer's voice information and convert the voice information into text information;

[0131] A matching unit 405, configured to match the text information with each of the historical search tags;

[0132] A second search unit 406, configured to use the answer information corresponding to the historical search tag as the answer information corresponding to the text information if the text information matches any historical search tag;

[0133] A third search unit 407, configured to extract each question tag in the text information and search for the answer information corresponding to each question tag in the filtered knowledge forest according to each question tag if the text information does not match all historical search tags.

[0134] In the device provided by the embodiment of the present invention, when a customer's incoming call is detected, the customer's telephone number and the incoming call area are obtained; and the knowledge forest is initially screened based on the incoming call area, specifically, reducing the subtrees and the branches on the subtrees in the knowledge forest. Based on the telephone number, each historical search tag of the customer in the historical service record is obtained. For example, all historical search tags of the questions consulted by the user within one month are obtained, and the specific time period involved in obtaining the historical search tags is not limited here. After the customer dials the incoming call, the customer is prompted to input voice information. When the customer's voice information is obtained, the voice information is converted into text information, and the text information is matched with each historical search tag. If the match is successful, the answer information corresponding to the historical search tag is used as the answer information of the text information. If the match fails, the answer information corresponding to each question tag in the text information is searched for in the filtered knowledge forest.

[0135] By applying the device provided by the embodiment of the present invention, answers to multiple questions can be quickly found through a voice question-and-answer method and fed back to the customer, improving the customer experience.

[0136] In the device provided by the embodiment of the present invention, the screening unit 402 includes:

[0137] An obtaining subunit, configured to obtain each subtree set in the knowledge forest, each subtree corresponding to a different service type, each subtree including multiple branches, and each branch corresponding to a service area;

[0138] A determining subunit, configured to determine each target service type supported by the incoming call area;

[0139] A screening subunit, configured to retain the subtrees corresponding to each target service type in the knowledge forest, and prune the branches of each subtree based on the incoming call area and the service areas corresponding to the branches in each subtree to obtain a filtered knowledge forest.

[0140] In the device provided by the embodiment of the present invention, the conversion unit 404 includes:

[0141] A noise reduction unit, configured to remove noise data in the voice information to obtain denoised voice information;

[0142] A conversion subunit, configured to convert the denoised voice information into text content and recognize the contextual meaning of the text content;

[0143] A correction subunit, configured to correct the text content based on the contextual meaning of the text content to obtain the text information corresponding to the voice information.

[0144] In the apparatus provided by an embodiment of the present invention, the third search unit 407 includes:

[0145] A segmentation subunit, configured to segment the context of the text information based on the contextual meaning of the text information;

[0146] An extraction subunit, configured to determine the service type to which each segment belongs and extract each keyword in each segment to obtain a keyword group for each segment;

[0147] A setting subunit, configured to set each of the keyword groups as a question tag;

[0148] A search subunit, configured to search, in the filtered knowledge forest, whether there is answer information corresponding to each question tag according to the service type corresponding to each question tag;

[0149] A feedback subunit, configured to, if there is answer information corresponding to each question tag, feedback each of the answer information to the customer;

[0150] A prompt subunit, configured to, if there is no answer information corresponding to any question tag, prompt the customer to transfer to the artificial seat service.

[0151] In the apparatus provided by an embodiment of the present invention, it further includes:

[0152] A text processing subunit, configured to delete stop words in the text information and perform synonym replacement on synonyms in the text information based on the contextual meaning of the text information.

[0153] For the specific working processes of each unit and subunit in the knowledge answer search apparatus disclosed in the above embodiments of the present invention, reference may be made to the corresponding content in the knowledge answer search method disclosed in the above embodiments of the present invention, which will not be elaborated here.

[0154] An embodiment of the present invention further provides a storage medium, where the storage medium includes stored instructions, and when the instructions run, the device where the storage medium is located is controlled to execute the above knowledge answer search method.

[0155] An embodiment of the present invention also provides an electronic device, and its structural schematic diagram is as Figure 5 shown, specifically including a memory 501, and one or more instructions 502, wherein one or more instructions 502 are stored in the memory 501 and are configured to be executed by one or more processors 503 to perform the following operations on the one or more instructions 502:

[0156] When it is detected that a customer calls in, obtain the customer's telephone number and the calling area;

[0157] Based on the calling area, perform a preliminary screening on a pre-set knowledge forest to obtain a screened knowledge forest;

[0158] Based on the telephone number, search for the historical service record of the telephone number, and extract each historical search tag in the historical service record;

[0159] Obtain the customer's voice information and convert the voice information into text information;

[0160] Match the text information with each of the historical search tags;

[0161] If the text information matches any historical search tag, use the answer information corresponding to the historical search tag as the answer information corresponding to the text information;

[0162] If the text information does not match all the historical search tags, extract each question tag in the text information, and search for the answer information corresponding to each question tag in the screened knowledge forest according to each question tag.

[0163] Each embodiment in this specification is described in a progressive manner. For the same or similar parts between each embodiment, reference can be made to each other. The key point of each embodiment is to illustrate the differences from other embodiments. In particular, for a system or a system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. The systems and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative work.

[0164] Those skilled in the art may further realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of both.

[0165] To clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0166] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for searching knowledge answers, characterized in that, it includes: When detecting that a customer calls in, obtain the customer's phone number and the calling area; Based on the calling area, conduct a preliminary screening on a pre-set knowledge forest to obtain a screened knowledge forest, including: obtaining each subtree set in the knowledge forest, each of the subtrees corresponding to a different business type, each of the subtrees containing multiple branches, each branch corresponding to a business area, each of the branches containing multiple leaves, each of the leaves corresponding to a knowledge question and answer information; determining each target business type supported by the calling area; retaining the subtrees in the knowledge forest corresponding to each of the target business types, and based on the calling area and the business areas corresponding to each branch in each of the subtrees, pruning each branch of each subtree to obtain a screened knowledge forest; Based on the phone number, search for the historical service records of the phone number and extract each historical search tag in the historical service records; the historical search tag is a keyword group of each keyword in the questions consulted by the customer in the historical service records; Obtain the voice information of the customer and convert the voice information into text information; there are multiple questions to be consulted in the text information; Match the text information with each of the historical search tags; If the text information matches any historical search tag, use the answer information corresponding to the historical search tag as the answer information corresponding to the text information; If the text information does not match all the historical search tags, extract each question tag in the text information and, according to each of the question tags, search for the answer information corresponding to each question tag in the screened knowledge forest; the question tag is a keyword group of each keyword in the questions consulted by the customer.

2. The method according to claim 1, characterized in that, the converting the voice information into text information includes: Removing the noise data in the voice information to obtain denoised voice information; Converting the denoised voice information into text content and identifying the context meaning of the text content; Based on the context meaning of the text content, correcting the text content to obtain the text information corresponding to the voice information.

3. The method according to claim 2, characterized in that, the extracting each question tag in the text information and, according to each of the question tags, searching for the answer information corresponding to each question tag in the screened knowledge forest includes: Based on the context meaning of the text information, segment the context of the text information; Determine the business type to which each segment belongs and extract each keyword in each segment to obtain a keyword group for each segment; Set each of the keyword groups as a question tag; In the screened knowledge forest, according to the business type corresponding to each question tag, search for whether there is answer information corresponding to each question tag; If there is answer information corresponding to each of the problem tags, then feedback each of the answer information to the customer; If there is no answer information corresponding to any of the problem tags, then prompt the customer to transfer to the artificial seat service.

4. The method according to claim 1 or 3, characterized in that, further comprising: Based on the context meaning of the text information, delete the stop words in the text information, and perform synonym replacement on the synonyms in the text information.

5. A knowledge answer search device, characterized in that, comprising: An acquisition unit, configured to acquire the customer's telephone number and the calling area when detecting that a customer calls in; A screening unit, configured to perform preliminary screening on a pre-set knowledge forest based on the calling area to obtain a screened knowledge forest; The screening unit includes: an acquisition subunit, a determination subunit, and a screening subunit; The acquisition subunit is configured to acquire each subtree set in the knowledge forest, each of the subtrees corresponds to a different service type, each of the subtrees includes multiple branches, each branch corresponds to a service area, and each branch includes multiple leaves, and each leaf corresponds to a knowledge question and answer information; The determination subunit is configured to determine each target service type supported by the calling area; The screening subunit is configured to retain the subtrees corresponding to each of the target service types in the knowledge forest, and prune the branches of each subtree based on the calling area and the service areas corresponding to the branches in each subtree to obtain a screened knowledge forest; A first search unit, configured to search for the historical service record of the telephone number based on the telephone number, and extract each historical search tag in the historical service record; the historical search tag is a keyword group of each keyword of the question consulted by the customer in the historical service record; A conversion unit, configured to acquire the voice information of the customer, and convert the voice information into text information; there are multiple questions to be consulted in the text information; A matching unit, configured to match the text information with each of the historical search tags; A second search unit, configured to, if the text information matches any of the historical search tags, use the answer information corresponding to the historical search tag as the answer information corresponding to the text information; A third search unit, configured to, if the text information does not match all of the historical search tags, extract each problem tag in the text information, and search for the answer information corresponding to each problem tag in the screened knowledge forest according to each problem tag; the problem tag is a keyword group of each keyword in the question consulted by the customer.

6. The device according to claim 5, characterized in that, The conversion unit includes: A denoising subunit, configured to remove the noise data in the voice information to obtain denoised voice information; A conversion subunit, configured to convert the denoised voice information into text content, and identify the context meaning of the text content; A correction subunit, configured to correct the text content based on the contextual meaning of the text content, so as to obtain the text information corresponding to the voice information.

7. The apparatus according to claim 6, wherein, the third search subunit includes: a segmentation subunit, configured to segment the context of the text information based on the contextual meaning of the text information; an extraction subunit, configured to determine the service type to which each segment belongs, and extract each keyword in each segment, so as to obtain a keyword group for each segment; a setting subunit, configured to set each keyword group as a question tag; a search subunit, configured to search in the filtered knowledge forest to find out whether there is answer information corresponding to each question tag according to the service type corresponding to each question tag; a feedback subunit, configured to, if there is answer information corresponding to each question tag, feedback each answer information to the customer; a prompt subunit, configured to, if there is no answer information corresponding to any question tag, prompt the customer to transfer to the artificial seat service.

8. The apparatus according to claim 7, wherein, it further includes: a text processing subunit, configured to delete stop words in the text information and perform synonym replacement on synonyms in the text information based on the contextual meaning of the text information.

Images