Search Method, Device, Electronic Device, Storage Medium and Program Product

Abstract

The present disclosure provides a retrieval method, which relates to the technical field of retrieval and can be applied to the financial field or other fields. The retrieval method includes: obtaining at least one target word to obtain a target word set; determining a first retrieval coefficient of each retrieval word in a preset retrieval word dictionary according to the target word set; constructing a retrieval word matrix according to the retrieval word dictionary and the determined first retrieval coefficients; obtaining a keyword matrix, where the keyword matrix includes a first keyword set of at least one audio file; constructing a target matrix according to the retrieval word matrix and the keyword matrix, where the target matrix includes at least one first keyword set and the retrieval score of this first keyword set generated according to the retrieval word matrix; and generating a retrieval result according to the target matrix. The present disclosure also provides a retrieval device, an electronic device, a storage medium, and a program product.

Description

Technical Field

[0001] The present disclosure relates to the field of retrieval technologies, and can be used in the financial field or any field other than the financial field. Specifically, the present disclosure relates to a retrieval method, apparatus, electronic device, storage medium, and program product. Background Art

[0002] In the financial field, a telephone banking system can provide customers with voice access and manual answering services, so that customer services can be handled and complaints and suggestions can be accepted by phone. The call records between customer service personnel and customers can be stored in the form of audio files. For these audio files, quality inspection personnel can listen to them to evaluate the audio files and assist customer service personnel in quickly mastering business skills. However, listening to a large number of audio files one by one is a very time-consuming task, with poor work efficiency and high costs. Summary of the Invention

[0003] In view of the above problems, the present disclosure provides a retrieval method, apparatus, electronic device, storage medium, and program product.

[0004] According to a first aspect of the present disclosure, there is provided a retrieval method, which includes:

[0005] Obtain at least one target word to obtain a target word set;

[0006] Determine a first retrieval coefficient of each retrieval word in a preset retrieval word dictionary according to the target word set;

[0007] Construct a retrieval word matrix according to the retrieval word dictionary and the determined first retrieval coefficients;

[0008] Obtain a keyword matrix, where the keyword matrix includes a first keyword set of at least one audio file;

[0009] Construct a target matrix according to the retrieval word matrix and the keyword matrix, where the target matrix includes at least one first keyword set and a retrieval score of the first keyword set generated according to the retrieval word matrix;

[0010] Generate a retrieval result according to the target matrix.

[0011] According to an embodiment of the present disclosure, the keyword matrix is constructed through the following steps:

[0012] Obtain at least one audio file;

[0013] Extract keywords from each audio file according to the retrieval word dictionary to obtain a first keyword set of each audio file;

[0014] The obtained first keyword set constitutes the keyword matrix.

[0015] According to an embodiment of the present disclosure, extracting keywords from each audio file according to a preset search term dictionary to obtain a first keyword set for each audio file includes:

[0016] Converting the audio file into a text file;

[0017] Extracting keywords that match the search terms in the search term dictionary from the text file to obtain a second keyword set;

[0018] Determining a second retrieval coefficient for each keyword in the second keyword set according to the number of occurrences of each keyword in the text file;

[0019] Constructing the first keyword set according to each keyword in the second keyword set and the second retrieval coefficient of each keyword.

[0020] According to an embodiment of the present disclosure, the keyword matrix is a matrix with multiple rows and multiple columns. The keyword matrix includes the first keyword sets of multiple audio files. One column of the keyword matrix includes a first keyword set. One row of the keyword matrix includes the second retrieval coefficients of multiple keywords with the same semantics. The retrieval term matrix is a matrix with one row and multiple columns. One column of the retrieval term matrix includes the first retrieval coefficient of a search term. Constructing a target matrix according to the retrieval term matrix and the keyword matrix includes:

[0021] Multiplying the retrieval term matrix by the keyword matrix to generate an intermediate matrix with one row and multiple columns. One column of the intermediate matrix includes the retrieval score of a first keyword set;

[0022] Storing the intermediate matrix as the last row into the keyword matrix to generate the target matrix, where each column of the target matrix includes at least one first keyword set and the retrieval score of the first keyword set.

[0023] According to an embodiment of the present disclosure, before the step of constructing a target matrix according to the retrieval term matrix and the keyword matrix, the retrieval method further includes:

[0024] Obtaining keywords of interest;

[0025] Removing a part of the keywords from the keyword matrix according to the keywords of interest to filter the keyword matrix;

[0026] Performing dimensionality reduction processing on the filtered keyword matrix using the singular value decomposition method.

[0027] According to an embodiment of the present disclosure, determining a first retrieval coefficient for each retrieval term in a preset retrieval term dictionary according to the target term set includes:

[0028] Removing a part of the retrieval terms from the retrieval term dictionary according to the concerned keyword to filter the retrieval term dictionary;

[0029] For at least one retrieval term in the filtered retrieval term dictionary,

[0030] Determining the first retrieval coefficient of the retrieval term according to whether there is a target term matching the retrieval term in the target term set.

[0031] According to an embodiment of the present disclosure, the retrieval term dictionary includes weights of at least one retrieval term. Determining the first retrieval coefficient of the retrieval term according to whether there is a target term matching the retrieval term in the target term set includes:

[0032] When there is a target term matching the retrieval term in the target term set, using the weight of the retrieval term as the first retrieval coefficient of the retrieval term;

[0033] When there is no target term matching the retrieval term in the target term set, setting the first retrieval coefficient of the retrieval term to zero.

[0034] According to an embodiment of the present disclosure, generating a retrieval result according to the target matrix includes:

[0035] Screening out at least a part of the first keyword sets in the target matrix according to the retrieval scores of each first keyword set in the target matrix;

[0036] Obtaining the audio files to which the screened first keyword sets belong;

[0037] Sorting the obtained audio files according to the retrieval scores to generate the retrieval result.

[0038] According to an embodiment of the present disclosure, the retrieval method further includes:

[0039] Obtaining newly added audio files;

[0040] Extracting keywords from the newly added audio files according to the retrieval term dictionary to obtain a third keyword set of the newly added audio files;

[0041] Updating the keyword matrix according to the third keyword set.

[0042] A second aspect of the present disclosure provides a retrieval device, which includes:

[0043] A first acquisition module, configured to acquire at least one target word to obtain a target word set;

[0044] A first retrieval coefficient determination module, configured to determine a first retrieval coefficient of each retrieval word in a preset retrieval word dictionary according to the target word set;

[0045] A retrieval word matrix construction module, configured to construct a retrieval word matrix according to the retrieval word dictionary and the determined first retrieval coefficients;

[0046] A second acquisition module, configured to acquire a keyword matrix, where the keyword matrix includes a first keyword set of at least one audio file;

[0047] A target matrix construction module, configured to construct a target matrix according to the retrieval word matrix and the keyword matrix, where the target matrix includes at least one first keyword set and a retrieval score of the first keyword set generated according to the retrieval word matrix;

[0048] A retrieval result generation module, configured to generate a retrieval result according to the target matrix.

[0049] A third aspect of the present disclosure provides an electronic device, including: one or more processors; a memory for storing one or more programs, where when the one or more programs are executed by the one or more processors, the one or more processors are caused to execute the above-mentioned retrieval method.

[0050] A fourth aspect of the present disclosure further provides a computer-readable storage medium, on which executable instructions are stored, and when the instructions are executed by a processor, the processor is caused to execute the above-mentioned retrieval method.

[0051] A fifth aspect of the present disclosure further provides a computer program product, including a computer program, and when the computer program is executed by a processor, the above-mentioned retrieval method is implemented.

[0052] The above one or more embodiments have the following advantages or beneficial effects:

[0053] After a user inputs a target word, the retrieval word matrix and the keyword matrix can be used to quickly screen out audio files corresponding to the target word, that is, this can effectively improve service quality, contribute to business handling, thereby enhancing the customer experience and reducing operating costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, the above content and other objects, features, and advantages of the present disclosure will become clearer. In the drawings:

[0055] Figure 1Schematically shows an application scenario diagram of a retrieval method, apparatus, electronic device, storage medium, and program product according to an embodiment of the present disclosure;

[0056] Figure 2 Schematically shows a flowchart of a retrieval method according to an embodiment of the present disclosure;

[0057] Figure 3 Schematically shows a flowchart of constructing a keyword matrix according to an embodiment of the present disclosure;

[0058] Figure 4 Schematically shows a flowchart of constructing a first keyword set according to an embodiment of the present disclosure;

[0059] Figure 5 Schematically shows a flowchart of constructing a target matrix according to an embodiment of the present disclosure;

[0060] Figure 6 Schematically shows a flowchart of filtering and dimensionality reduction of a keyword matrix according to an embodiment of the present disclosure;

[0061] Figure 7 Schematically shows one of the flowcharts of determining a first retrieval coefficient according to an embodiment of the present disclosure;

[0062] Figure 8 Schematically shows another flowchart of determining a first retrieval coefficient according to an embodiment of the present disclosure;

[0063] Figure 9 Schematically shows a flowchart of generating a retrieval result according to an embodiment of the present disclosure;

[0064] Figure 10 Schematically shows a flowchart of updating a keyword matrix according to an embodiment of the present disclosure;

[0065] Figure 11 Schematically shows a structural block diagram of a retrieval apparatus according to an embodiment of the present disclosure;

[0066] Figure 12 Schematically shows a block diagram of an electronic device suitable for implementing a retrieval method according to an embodiment of the present disclosure. Detailed implementation manners

[0067] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present disclosure. In the following detailed description, for the sake of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments can also be implemented without these specific details. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily confusing the concepts of the present disclosure.

[0068] The terms used herein are merely for describing specific embodiments and are not intended to limit the present disclosure. The terms "including", "comprising", etc. used herein indicate the presence of the described features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0069] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those of ordinary skill in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification, and should not be interpreted in an idealized or overly rigid manner.

[0070] In the case of using expressions such as "at least one of A, B, and C, etc.", generally, it should be interpreted according to the meaning commonly understood by those of ordinary skill in the art (for example, "a system having at least one of A, B, and C" should include, but not be limited to, a system having only A, only B, only C, having A and B, having A and C, having B and C, and / or having A, B, and C, etc.).

[0071] It should be noted that the retrieval method, device, electronic device, storage medium, and program product provided by the present disclosure relate to the field of retrieval technology. The retrieval method, device, electronic device, storage medium, and program product provided by the embodiments of the present disclosure can be applied to retrieval services in the financial field or any field outside the financial field. The present disclosure does not limit the application fields of the retrieval method, device, electronic device, storage medium, and program product.

[0072] In the technical solution of the present disclosure, the processing of collection, storage, use, processing, transmission, provision, disclosure, and application, etc. of the user's personal information involved all comply with the provisions of relevant laws and regulations, have taken necessary confidentiality measures, and do not violate public order and good customs.

[0073] The embodiments of the present disclosure provide a retrieval method, which includes: obtaining at least one target word to obtain a target word set. According to the target word set, determining a first retrieval coefficient for each retrieval word in a preset retrieval word dictionary. Constructing a retrieval word matrix according to the retrieval word dictionary and the determined first retrieval coefficients. Obtaining a keyword matrix, where the keyword matrix includes a first keyword set of at least one audio file. Constructing a target matrix according to the retrieval word matrix and the keyword matrix, where the target matrix includes at least one first keyword set and the retrieval score of the first keyword set generated according to the retrieval word matrix. Generating a retrieval result according to the target matrix.

[0074] In the embodiments of the present disclosure, after a user inputs a target word, the retrieval word matrix and the keyword matrix can be used to quickly screen out the audio files corresponding to the target word. That is, this can effectively improve the service quality, contribute to business handling, thereby enhancing the customer experience and reducing the operation cost.

[0075] Figure 1 Schematically shows an application scenario diagram of a retrieval method, device, electronic device, storage medium, and program product according to an embodiment of the present disclosure, as Figure 1 shown. According to the application scenario 100 of this embodiment, it may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, etc.

[0076] Users can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications may be installed on the terminal devices 101, 102, 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only as examples).

[0077] The terminal devices 101, 102, 103 may be various electronic devices with a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop portable computers, and desktop computers, etc.

[0078] The server 105 may be a server providing various services, such as a background management server (only as an example) that supports the websites browsed by users using the terminal devices 101, 102, 103. The background management server may analyze and process data such as user requests received, and feedback the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0079] It should be noted that the retrieval method provided by the embodiments of the present disclosure can generally be executed by the server 105. Correspondingly, the retrieval device provided by the embodiments of the present disclosure can generally be set in the server 105. The retrieval method provided by the embodiments of the present disclosure can also be executed by a server or a server cluster different from the server 105 and capable of communicating with the terminal devices 101, 102, 103 and / or the server 105. Correspondingly, the retrieval device provided by the embodiments of the present disclosure can also be set in a server or a server cluster different from the server 105 and capable of communicating with the terminal devices 101, 102, 103 and / or the server 105.

[0080] It should be understood,Figure 1 The number of terminal devices, networks, and servers in [the above] is merely illustrative. According to the implementation requirements, there can be any number of terminal devices, networks, and servers.

[0081] Based on the Figure 1 scenario described below, through Figure 2 a retrieval method for public embodiments will be described in detail.

[0082] Figure 2 Schematically shows a flowchart of a retrieval method according to an embodiment of the present disclosure. As Figure 2 shown, the retrieval method of this embodiment includes steps S210 to step S260. It should be noted that although Figure 2 the steps in [the figure] are shown in sequence according to the indication of the arrow, however, these steps are not necessarily executed in the order indicated by the arrow. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and they can be executed in other orders. Moreover, at least some of the steps in the figure may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. Their execution order is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or sub-steps or stages of other steps.

[0083] In step S210, obtain at least one target word to obtain a target word set.

[0084] In the embodiments of the present disclosure, the retrieval method can be applied to the retrieval service of telephone recordings in the financial field. Staff can retrieve the required telephone recordings through the retrieval method of the embodiments of the present disclosure. The target word can refer to the retrieval target, and the obtained target word can be one or more. For example, the target word can include "complaint". At this time, telephone recordings involving "complaint" can be retrieved from the telephone recording library. The target word can also include "complaint" and "credit card". At this time, telephone recordings involving "complaint" and "credit card" can be retrieved from the telephone recording library.

[0085] In step S220, according to the target word set, determine the first retrieval coefficient of each retrieval word in the preset retrieval word dictionary.

[0086] In the embodiments of the present disclosure, the retrieval word dictionary may include popular keywords related to the business to which the telephone recording belongs. For example, when the telephone recording belongs to a bank customer service, the retrieval word dictionary may include "complaint", "credit card", "suggestion", "debit card", and so on. Optionally, the first retrieval coefficient of each retrieval word in the retrieval word dictionary may be determined according to whether the target word contains the retrieval word in the retrieval word dictionary. For example, when the target word contains the retrieval word, a non-zero value may be assigned to the retrieval word as the first retrieval coefficient. When the target word does not contain the retrieval word, the first retrieval coefficient of the retrieval word may be set to zero.

[0087] In step S230, a retrieval word matrix is constructed according to the retrieval word dictionary and the determined first retrieval coefficient.

[0088] In the embodiments of the present disclosure, the retrieval word matrix may be an m*n-dimensional matrix. Wherein, m represents the number of rows of the elements in the retrieval word matrix, and n represents the number of columns of the elements in the retrieval word matrix. Both m and n are positive integers.

[0089] Optionally, m = 1, that is, the retrieval word matrix is a matrix including one row and multiple columns of elements.

[0090] Optionally, each element in the retrieval word matrix may include the first retrieval coefficient of a retrieval word, and the retrieval words in different elements are different.

[0091] In step S240, a keyword matrix is obtained. The keyword matrix includes the first keyword sets of at least one audio file.

[0092] In the embodiments of the present disclosure, the keyword matrix may be pre-generated or may be generated in real time each time a retrieval is performed. The keyword matrix may be an x*y-dimensional matrix, where x represents the number of rows of the elements in the keyword matrix, n is the number of columns of the elements in the keyword matrix, and both x and y are positive integers.

[0093] Optionally, x = n, y = a, and a may represent the number of audio files in the audio library, that is, the number of rows of the elements in the keyword matrix is equal to the number of columns of the elements in the retrieval word matrix, and the number of columns of the elements in the keyword matrix is equal to the number of audio files.

[0094] Optionally, one column in the keyword matrix includes the first keyword set of an audio file, and the audio files in different columns are different.

[0095] Optionally, the first keyword set includes multiple keywords. The keywords in the first keyword set may include: the keywords selected from the audio file and matching the retrieval word according to the retrieval word in the retrieval word dictionary. For one column of elements in the keyword matrix, each element may include a keyword, and the keywords in different elements are different.

[0096] Optionally, for a column of elements in the keyword matrix, the elements in this column can correspond one-to-one with the elements in the retrieval term matrix. For example, the keyword of the i-th row element in the keyword matrix and the retrieval term of the i-th column element in the retrieval term matrix are the same "word".

[0097] In step S250, according to the retrieval term matrix and the keyword matrix, a target matrix is constructed. The target matrix includes at least one first keyword set and the retrieval score of this first keyword set generated according to the retrieval term matrix.

[0098] In the embodiments of the present disclosure, the retrieval-side matrix can be multiplied by the keyword matrix, and the product of the two is a new matrix. Each element in this matrix can include the retrieval score of a first keyword set. Then, according to this product, a target matrix is generated. For example, the product of the retrieval-side matrix and the keyword matrix can be stored in the keyword matrix to obtain the target matrix. In this way, each column in the target matrix includes a first keyword set of an audio file and the retrieval score of this first keyword set.

[0099] In step S260, according to the target matrix, a retrieval result is generated.

[0100] In the embodiments of the present disclosure, each column in the target matrix includes a first keyword set of an audio file and the retrieval score of this first keyword set. Therefore, the first keyword sets with higher retrieval scores can be extracted, and the audio files to which these first keyword sets belong can be pushed to the user as the retrieval result.

[0101] The above method can be applied in a telephone banking system. Through the retrieval method of the embodiments of the present disclosure, after the user inputs the target word, the retrieval term matrix and the keyword matrix can be used to quickly screen out the audio files corresponding to the target word, that is, screen out the recordings required by the user. This can effectively improve the service quality, contribute to business handling, thereby enhancing the customer experience and reducing the operating cost.

[0102] The following combines Figures 2 to 10 to further illustrate the retrieval method of the embodiments of the present disclosure.

[0103] Figure 3 Schematically shows a flowchart of constructing a keyword matrix according to an embodiment of the present disclosure. As Figure 3 shown, in some specific embodiments, the keyword matrix is constructed through steps S310 to S330.

[0104] In step S310, at least one audio file is obtained.

[0105] In step S320, according to the search term dictionary, keywords are extracted from each audio file to obtain a first keyword set for each audio file.

[0106] In the embodiments of the present disclosure, according to the search terms in the search term dictionary, keywords matching the search terms can be extracted from the audio file, and then a first keyword set is constructed based on the extracted keywords.

[0107] Optionally, the keywords matching the search terms may include keywords having the same semantics as the search terms or keywords that are the same as the search terms. For example, if the search terms in the search term dictionary include "complaint", "suggestion", "credit card", and "debit card", then words having the same semantics as "complaint", "suggestion", "credit card", and "debit card" can be extracted from the audio file as the above-mentioned keywords to construct a first keyword set; or, directly extract the words "complaint", "suggestion", "credit card", and "debit card" from the audio file as keywords to construct a first keyword set. For example, the audio file may include: "Hello, what business do you handle; Hello, I want to complain about the customer service staff who just cancelled my debit card for me; What's your card number, my card number is XXXX; Wait a moment, I'll check the business staff who just handled the debit card for you. What do you want to complain about him?" For the above audio file, it can be compared with the search terms in the search term dictionary, and "complaint" and "debit card" can be selected from it to construct the first keyword set of this audio file.

[0108] Figure 4 Schematically shows a flowchart of constructing a first keyword set according to an embodiment of the present disclosure, as Figure 4 shown. In some specific embodiments, step S320 includes steps S321 to S324.

[0109] In step S321, the audio file is converted into a text file.

[0110] In step S322, keywords matching the search terms in the search term dictionary are extracted from the text file to obtain a second keyword set.

[0111] In the embodiments of the present disclosure, the audio file can be converted into an initial text file through semantic recognition and other means. After that, through word segmentation, cleaning and other means to remove useless words, the text file of each audio file is obtained. Then, from each text file, keywords having the same semantics as the search terms or keywords that are the same as the search terms are extracted to obtain a second keyword set.

[0112] In step S323, according to the number of occurrences of each keyword in the second keyword set in the text file, determine the second retrieval coefficient of each keyword in the second keyword set.

[0113] In one example, the number of occurrences of each keyword in the text file can be directly used as the second retrieval coefficient of the keyword.

[0114] In another example, according to the number of occurrences of the keyword in the text file, combined with other attributes of each keyword, the second retrieval coefficient of the keyword can be determined. For example, based on the number of occurrences of the keyword in the text file, taking the importance of the keyword as the adjustment coefficient, the adjusted number of occurrences can be used as the second retrieval coefficient of the keyword. In this way, the second retrieval coefficient of important keywords with fewer occurrences can be appropriately increased, and the second retrieval coefficient of unimportant keywords with more occurrences can be appropriately decreased.

[0115] In step S324, construct the first keyword set according to each keyword in the second keyword set and the second retrieval coefficient of each keyword.

[0116] In the embodiments of the present disclosure, the first keyword set may include a matrix with multiple rows in one column, and each element in the matrix includes the second retrieval coefficient of a keyword, and the keywords in different elements are different.

[0117] Optionally, for different first keyword sets, the keywords included in the elements in the same row are the same. For example, if the keyword included in the j-th row element of the first keyword set A is the second retrieval coefficient of "complaint", then the keyword included in the j-th row element of the first keyword set A' is also the second retrieval coefficient of "complaint". It should be noted that the first keyword set A belongs to the audio file B, and the second keyword set A' belongs to the audio file B'. When the number of occurrences of "complaint" in the audio file B is different from the number of occurrences of "complaint" in the audio file B', the second retrieval coefficient of the j-th row element in the first keyword set A is also different from the second retrieval coefficient of the j-th row element in the first keyword set A'.

[0118] Exemplarily, the number of occurrences of "complaint" in the audio file B1 is "1", the number of occurrences of "suggestion" in the audio file B1 is "0", the number of occurrences of "credit card" in the audio file B1 is "0", and the number of occurrences of "debit card" in the audio file B1 is "7". Then, the first keyword set A1 can be obtained.

[0119]

[0120] The occurrence times of "complaint" in audio file B2 is "1", the occurrence times of "suggestion" in audio file B2 is "1", the occurrence times of "credit card" in audio file B2 is "0", and the occurrence times of "debit card" in audio file B2 is "4". Then, the first keyword set A2 can be obtained.

[0121]

[0122] The occurrence times of "complaint" in audio file B3 is "0", the occurrence times of "suggestion" in audio file B3 is "1", the occurrence times of "credit card" in audio file B3 is "0", and the occurrence times of "debit card" in audio file B3 is "0". Then, the first keyword set A3 can be obtained.

[0123]

[0124] After step S320, step S330 is executed.

[0125] In step S330, a keyword matrix is formed by the obtained first keyword sets.

[0126] In the embodiment of the present disclosure, multiple first keyword sets with one column and multiple rows can be merged into a matrix with multiple rows and multiple columns to obtain a keyword matrix. For example, after merging the above A1, A2, and A3, the keyword matrix C1 is obtained.

[0127]

[0128] In some specific embodiments, the keyword matrix is a matrix with multiple rows and multiple columns. The keyword matrix includes the first keyword sets of multiple audio files. One column of the keyword matrix includes one first keyword set. One row of the keyword matrix includes the second retrieval coefficients of multiple keywords with the same semantics. The retrieval word matrix is a matrix with one row and multiple columns. One column of the retrieval word matrix includes the first retrieval coefficient of one retrieval word.

[0129] In the embodiment of the present disclosure, the keyword matrix can be as shown in the keyword matrix C1 in the previous text, and the retrieval word matrix can be as shown in the retrieval word matrix D1 in the previous text, so it will not be elaborated here.

[0130] In the embodiment of the present disclosure, as described in the previous text, the retrieval word dictionary can include "complaint", "suggestion", "credit card", and "debit card". The target word can include "complaint". At this time, a non-zero value can be assigned to "complaint" as the first retrieval coefficient of "complaint", and the first retrieval coefficients of "suggestion", "credit card", and "debit card" are set to zero to obtain the retrieval word matrix D1.

[0131] D1 = [1 0 0 0]

[0132] Figure 5 Schematically shows a flowchart of constructing a target matrix according to an embodiment of the present disclosure, as Figure 5 shown, step S250 includes step S251 and step S252.

[0133] In step S251, multiply the retrieval term matrix by the keyword matrix to generate an intermediate matrix with one row and multiple columns, where one column in the intermediate matrix includes the retrieval scores of a first keyword set.

[0134] In the embodiment of the present disclosure, the intermediate matrix is a matrix with one row and multiple columns, and the number of columns in the intermediate matrix is the same as the number of columns in the keyword matrix.

[0135] For example, multiply the retrieval matrix D1 by the keyword matrix C1 to obtain the intermediate matrix E1. Specifically, the retrieval matrix D1 can be multiplied by each column in the keyword matrix C1 respectively and summed to obtain the retrieval scores of each first keyword set.

[0136] E1 = [8 5 0]

[0137] Among them, in the intermediate matrix E1, "8" is the retrieval score of the first keyword set A1, "5" is the retrieval score of the first keyword set A2, and "0" is the retrieval score of the first keyword set A3.

[0138] In step S252, store the intermediate matrix as the last row into the keyword matrix to generate the target matrix, where each column in the target matrix includes at least one first keyword set and the retrieval score of this first keyword set.

[0139] For example, store the intermediate matrix E1 into the last row of the keyword matrix C1 to obtain the target matrix F.

[0140]

[0141] Among them, in the target matrix F1, the first column includes the first keyword set A1 and its retrieval score, the second column includes the first keyword set A2 and its retrieval score, and the third column includes the first keyword set A3 and its retrieval score.

[0142] In some specific embodiments, the above-mentioned retrieval matrix and keyword matrix can be further optimized to improve the retrieval efficiency.

[0143] Figure 6 Schematically shows a flowchart of filtering and dimensionality reduction of the keyword matrix according to an embodiment of the present disclosure, as Figure 6 shown, before step S250, the retrieval method further includes steps S410 to S430.

[0144] In step S410, obtain the keywords of interest.

[0145] In the embodiments of the present disclosure, the keywords of interest can be the key words required for the current retrieval. For example, the keywords of interest may include "complaint", "credit card", and "debit card", that is to say, in this retrieval process, "suggestion" in the retrieval dictionary can be ignored.

[0146] In step S420, according to the keywords of interest, remove a part of the keywords from the keyword matrix to filter the keyword matrix.

[0147] As described above, in the keyword matrix, the elements in the same row can include the same keyword. For example, for the keyword matrix C1, the elements in the first row all include the keyword "complaint", the elements in the second row all include the keyword "suggestion", the elements in the third row all include the keyword "credit card", and the elements in the fourth row all include the keyword "debit card". Therefore, by comparing the keywords of interest with the retrieval word dictionary, the words to be filtered out can be determined, and then the elements corresponding to the filtered-out words can be removed from the keyword matrix to filter the keyword matrix.

[0148] For example, the keywords of interest include "complaint", "credit card", and "debit card". After filtering the keyword matrix C1, the filtered keyword matrix C2 is obtained.

[0149]

[0150] In step S430, use the singular value decomposition method to perform dimensionality reduction on the filtered keyword matrix.

[0151] In the embodiments of the present disclosure, the dimensionality of the keyword matrix can be reduced by the singular value decomposition method. Assume that a given matrix X of dimension a*b with rank r, then the singular value decomposition form of X is

[0152]

[0153] where U is an a*a dimensional orthogonal matrix, V is a b*b dimensional orthogonal matrix, D r*r is an r*r dimensional diagonal matrix, and its diagonal elements are the non-zero singular values σ z of matrix X, and they are arranged in non-increasing order, that is, σ z ≥σ z ≥…≥σ z , and 0 is a zero element matrix. The rank of matrix X is r. Removing the zero singular values of X from equation (1), the reduced form of the singular value decomposition of X is obtained, as shown in formula (2).

[0154]

[0155] where, uz is the z-th inner vector of matrix U, v z is the z-th inner vector of V. Remove σ z values that are very small or zero to complete the singular value decomposition of the matrix.

[0156] In step S250, the retrieval term matrix can be multiplied by the keyword matrix after singular value decomposition, and the product can be stored in the last row of the keyword matrix to obtain the target matrix. In this way, the dimensionality of the keyword matrix can be reduced after singular value decomposition, thus greatly saving computing resources and improving retrieval efficiency.

[0157] Figure 7 Schematically shows one of the flowcharts for determining the first retrieval coefficient according to an embodiment of the present disclosure. As Figure 7 shown, in some specific embodiments, step S220 includes step S221 and step S222.

[0158] In step S221, according to the keywords of interest, a part of the retrieval terms are removed from the retrieval term dictionary to filter the retrieval term dictionary.

[0159] As described above, the retrieval term dictionary may include "complaint", "suggestion", "credit card", and "debit card", and the keywords of interest may include "complaint", "credit card", and "debit card". Therefore, "suggestion" in the retrieval dictionary can be removed, and the remaining retrieval terms form the filtered retrieval term dictionary.

[0160] In step S222, for at least one retrieval term in the filtered retrieval term dictionary, the first retrieval coefficient of the retrieval term is determined according to whether there is a target term in the target term set that matches the retrieval term.

[0161] For example, the first retrieval coefficient of the retrieval term can be determined according to whether there is a target term with the same semantics as the retrieval term in the target term set. Or, the first retrieval coefficient of the retrieval term can be determined according to whether there is a target term that is the same as the retrieval term in the target term set.

[0162] For example, the target term set includes "complaint", and the filtered retrieval term dictionary includes "complaint", "credit card", and "debit card". At this time, the retrieval term matrix D2 can be obtained.

[0163] D1 = [1 0 0]

[0164] In some specific embodiments, the search term dictionary includes the weights of at least one search term. In the embodiments of the present disclosure, the weights of the search terms can be determined according to actual needs. For example, the weight of "complaint" can be set to "6", the weight of "suggestion" can be set to "3", the weight of "credit card" can be set to "2", and the weight of "debit card" can be set to "1".

[0165] Figure 8 Schematically shows the second flowchart of determining the first search coefficient according to the embodiments of the present disclosure, as Figure 8 shown, step S222 includes step S2221 to step S2223.

[0166] In step S2221, it is judged whether there is a target word in the target word set that matches the search term. If so, step S2232 is executed; if not, step S2233 is executed.

[0167] In step S2222, the weight of the search term is used as the first search coefficient of the search term.

[0168] In step S2223, the first search coefficient of the search term is set to zero.

[0169] For example, the target word set includes "complaint" and "debit card", and the filtered search term dictionary includes "complaint", "credit card" and "debit card", where the weights of each search term are as described above, which are "6", "2" and "1" respectively. At this time, the search term matrix D3 can be obtained.

[0170] D3 = [6 0 1]

[0171] Exemplarily, it is assumed that the keyword matrix after singular value decomposition is C3.

[0172]

[0173] In step S250, the search matrix D3 can be multiplied by the keyword matrix C3 to obtain the intermediate matrix E2.

[0174] E2 = [12 9 0]

[0175] In the intermediate matrix E2, "12" is the search score of the first keyword set A1', "9" is the search score of the first keyword set A2', and "0" is the search score of the first keyword set A3'.

[0176] The intermediate matrix E2 is stored as the last row in the keyword matrix C3 to obtain the target matrix F2.

[0177]

[0178] Among them, in the target matrix F2, the first column includes the first keyword set A1' and its retrieval score "12", the second column includes the first keyword set A2' and its retrieval score "9", and the third column includes the first keyword set A3' and its retrieval score "0".

[0179] Figure 9 Schematically shows a flowchart of generating retrieval results according to an embodiment of the present disclosure, as Figure 9 shown, in some specific embodiments, step S260 includes steps S261 to S263.

[0180] In step S261, at least a part is screened out from the first keyword sets in the target matrix according to the retrieval scores of each first keyword set in the target matrix.

[0181] In the embodiment of the present disclosure, the first keyword sets with larger retrieval scores can be screened out from the target matrix. For example, the first keyword sets with retrieval scores greater than 3 can be screened out. Taking the target matrix F2, the first keyword sets A1' and A2' with retrieval scores greater than 3 can be obtained by screening.

[0182] In step S262, obtain the audio files to which the screened first keyword sets belong.

[0183] For example, in the embodiment of the present disclosure, the audio file B1' to which the first keyword set A1' belongs and the audio file B2' to which the first keyword set A2' belongs can be obtained.

[0184] In step S263, sort the obtained audio files according to the retrieval scores to generate retrieval results.

[0185] For example, in the embodiment of the present disclosure, the audio file B1' can be ranked first and the audio file B2' can be ranked second to generate retrieval results and push them to the user.

[0186] Figure 10 Schematically shows a flowchart of updating a keyword matrix according to an embodiment of the present disclosure, as Figure 10 shown, in some specific embodiments, the retrieval method further includes steps S510 to S530.

[0187] In step S510, obtain the newly added audio files.

[0188] In step S520, extract keywords from the newly added audio files according to the retrieval word dictionary to obtain the third keyword set of the newly added audio files.

[0189] In step S530, update the keyword matrix according to the third keyword set.

[0190] In the embodiments of the present disclosure, the update of the keyword matrix may include manual update or automatic update, which can be specifically determined according to actual needs and will not be limited herein. Through steps S510 to S530, the keyword matrix can be updated in real time, and in subsequent retrieval processes, the updated keyword matrix is used to participate in the retrieval to improve the retrieval effectiveness.

[0191] In the embodiments of the present disclosure, after the user inputs a target word, the retrieval word matrix and the keyword matrix can be used for retrieval to obtain the required audio file, so that it is no longer necessary to listen to each file one by one, effectively improving the retrieval efficiency.

[0192] Based on the above retrieval method, the present disclosure also provides a retrieval device. The following will be combined with Figure 11 to describe this device in detail.

[0193] Figure 11 Schematically shows a structural block diagram of a retrieval device according to an embodiment of the present disclosure. As Figure 11 shown, the retrieval device 1100 of this embodiment includes a first acquisition module 1110, a first retrieval coefficient determination module 1120, a retrieval word matrix construction module 1130, a second acquisition module 1140, a target matrix construction module 1150, and a retrieval result generation module 1160.

[0194] The first acquisition module 1110 is configured to acquire at least one target word to obtain a target word set. In one embodiment, the first acquisition module 1110 may be configured to execute step S210 described above, which will not be elaborated herein.

[0195] The first retrieval coefficient determination module 1120 is configured to determine a first retrieval coefficient of each retrieval word in a preset retrieval word dictionary according to the target word set. In one embodiment, the first retrieval coefficient determination module 1120 may be configured to execute step S220 described above, which will not be elaborated herein.

[0196] The retrieval word matrix construction module 1130 is configured to construct a retrieval word matrix according to the retrieval word dictionary and the determined first retrieval coefficients. In one embodiment, the retrieval word matrix construction module 1130 may be configured to execute step S230 described above, which will not be elaborated herein.

[0197] The second acquisition module 1140 is configured to acquire a keyword matrix, and the keyword matrix includes a first keyword set of at least one audio file. In one embodiment, the second acquisition module 1140 may be configured to execute step S240 described above, which will not be elaborated herein.

[0198] The target matrix construction module 1150 is configured to construct a target matrix according to the retrieval term matrix and the keyword matrix. The target matrix includes at least one first keyword set and the retrieval scores of this first keyword set generated according to the retrieval term matrix. In one embodiment, the target matrix construction module 1150 may be configured to execute step S250 described above, which will not be elaborated here.

[0199] The retrieval result generation module 1160 is configured to generate a retrieval result according to the target matrix. In one embodiment, the retrieval result generation module 1160 may be configured to execute step S260 described above, which will not be elaborated here.

[0200] The above device can be applied in a telephone banking system. Through the retrieval device of the embodiments of the present disclosure, after a user inputs a target word, the retrieval term matrix and the keyword matrix can be used to quickly screen out the audio files corresponding to the target word, that is, screen out the recordings required by the user, which can effectively improve the service quality, contribute to business handling, thereby enhancing the customer experience and reducing the operation cost.

[0201] According to an embodiment of the present disclosure, any multiple modules among the first acquisition module 1110, the first retrieval coefficient determination module 1120, the retrieval term matrix construction module 1130, the second acquisition module 1140, the target matrix construction module 1150, and the retrieval result generation module 1160 may be combined and implemented in one module, or any one of them may be split into multiple modules. Or, at least part of the functions of one or more of these modules may be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the first acquisition module 1110, the first retrieval coefficient determination module 1120, the retrieval term matrix construction module 1130, the second acquisition module 1140, the target matrix construction module 1150, and the retrieval result generation module 1160 may be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on chip, a system on substrate, a system on package, an application specific integrated circuit (ASIC), or may be implemented by any other reasonable means such as hardware or firmware through circuit integration or packaging, or may be implemented in any one of the three implementation manners of software, hardware, and firmware or in an appropriate combination of any several of them. Or, at least one of the first acquisition module 1110, the first retrieval coefficient determination module 1120, the retrieval term matrix construction module 1130, the second acquisition module 1140, the target matrix construction module 1150, and the retrieval result generation module 1160 may be at least partially implemented as a computer program module, and when the computer program module runs, it can execute the corresponding functions.

[0202] Figure 12A block diagram of an electronic device suitable for implementing a retrieval method according to an embodiment of the present disclosure is schematically shown, as Figure 12 shown, the electronic device 1200 according to an embodiment of the present disclosure includes a processor 1201, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1202 or a program loaded from a storage section 1208 into a random access memory (RAM) 1203. The processor 1201 can include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor, and / or a related chipset, and / or a dedicated microprocessor (such as an application-specific integrated circuit (ASIC)), etc. The processor 1201 can also include on-board memory for caching purposes. The processor 1201 can include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0203] In the RAM 1203, various programs and data required for the operation of the electronic device 1200 are stored. The processor 1201, the ROM 1202, and the RAM 1203 are connected to each other via a bus 1204. The processor 1201 executes various operations of the method flow according to an embodiment of the present disclosure by executing the programs in the ROM 1202 and / or the RAM 1203. It should be noted that the program can also be stored in one or more memories other than the ROM 1202 and the RAM 1203. The processor 1201 can also execute various operations of the method flow according to an embodiment of the present disclosure by executing the programs stored in one or more memories.

[0204] According to an embodiment of the present disclosure, the electronic device 1200 can further include an input / output (I / O) interface 1205, and the input / output (I / O) interface 1205 is also connected to the bus 1204. The electronic device 1200 can further include one or more of the following components connected to the I / O interface 1205: an input section 1206 including a keyboard, a mouse, etc. An output section 1207 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc. A storage section 1208 including a hard disk, etc.; and a communication section 1209 including a network interface card such as a LAN card, a modem, etc. The communication section 1209 performs communication processing via a network such as the Internet. A drive 1210 is also connected to the I / O interface 1205 as needed. A removable medium 1211, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1210 as needed so that a computer program read from it can be installed into the storage section 1208 as needed.

[0205] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist alone without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the methods according to the embodiments of the present disclosure are implemented.

[0206] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, may include but is not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In the present disclosure, the computer-readable storage medium may be any tangible medium that contains or stores a program, and the program may be used by or in combination with an instruction execution system, apparatus, or device. For example, according to an embodiment of the present disclosure, the computer-readable storage medium may include the above-described ROM 1202 and / or RAM 1203 and / or one or more memories other than ROM 1202 and RAM 1203.

[0207] Embodiments of the present disclosure also include a computer program product, which includes a computer program that contains program code for executing the method shown in the flowchart. When the computer program product runs in a computer system, the program code is used to cause the computer system to implement the item recommendation method provided by the embodiments of the present disclosure.

[0208] When the computer program is executed by the processor 1201, the above functions defined in the system / apparatus of the embodiments of the present disclosure are executed. According to an embodiment of the present disclosure, the above-described systems, apparatuses, modules, units, etc. may be implemented by computer program modules.

[0209] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices and magnetic storage devices. In another embodiment, the computer program may also be transmitted and distributed in the form of a signal on a network medium, and is downloaded and installed through the communication part 1209, and / or installed from the removable medium 1211. The program code included in the computer program may be transmitted by any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0210] In such an embodiment, the computer program can be downloaded and installed from a network through the communication section 1209, and / or installed from the removable medium 1211. When the computer program is executed by the processor 1201, the above-described functions defined in the system of the embodiments of the present disclosure are performed. According to an embodiment of the present disclosure, the above-described systems, devices, apparatuses, modules, units, etc. can be implemented by computer program modules.

[0211] According to an embodiment of the present disclosure, program code for executing the computer program provided by the embodiments of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level procedures and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages include, but are not limited to, such as Java, C++, python, the "C" language, or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, by connecting through the Internet using an Internet service provider).

[0212] 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 disclosure. In this regard, each block in the flowchart or block diagram may 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 may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may 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 combinations 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.

[0213] Those skilled in the art can understand that the features recited in the various embodiments and / or claims of the present disclosure can be combined or combined in various ways, even if such combinations or combinations are not explicitly recited in the present disclosure. In particular, without departing from the spirit and teachings of the present disclosure, the features recited in the various embodiments and / or claims of the present disclosure can be combined and combined in various ways. All such combinations and / or combinations fall within the scope of the present disclosure.

[0214] The embodiments of the present disclosure have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of the present disclosure. Although the embodiments have been described separately above, this does not mean that the measures in each embodiment cannot be used advantageously in combination. The scope of the present disclosure is defined by the appended claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art can make various substitutions and modifications, and all such substitutions and modifications should fall within the scope of the present disclosure.

Claims

1. A retrieval method, characterized in that, comprising: obtaining at least one target word to obtain a target word set; determining a first retrieval coefficient for each retrieval word in a preset retrieval word dictionary according to the target word set; constructing a retrieval word matrix according to the retrieval word dictionary and the determined first retrieval coefficients; obtaining a keyword matrix, where the keyword matrix includes a first keyword set of at least one audio file; constructing a target matrix according to the retrieval word matrix and the keyword matrix, where the target matrix includes at least one first keyword set and the retrieval score of this first keyword set generated according to the retrieval word matrix; generating a retrieval result according to the target matrix; The keyword matrix is constructed through the following steps: obtaining at least one audio file; extracting keywords from each audio file according to the retrieval word dictionary to obtain a first keyword set of each audio file; constituting the keyword matrix by the obtained first keyword sets; The extracting keywords from each audio file according to the retrieval word dictionary to obtain a first keyword set of each audio file includes: converting the audio file into a text file; extracting keywords matching the retrieval words in the retrieval word dictionary from the text file to obtain a second keyword set; determining a second retrieval coefficient for each keyword in the second keyword set according to the number of occurrences of each keyword in the text file; constructing the first keyword set according to each keyword in the second keyword set and the second retrieval coefficient of each keyword; The keyword matrix is a matrix with multiple rows and multiple columns. The keyword matrix includes first keyword sets of multiple audio files. One column of the keyword matrix includes a first keyword set. One row of the keyword matrix includes the second retrieval coefficients of multiple keywords with the same semantics. The retrieval word matrix is a matrix with one row and multiple columns. One column of the retrieval word matrix includes the first retrieval coefficient of a retrieval word. The constructing the target matrix according to the retrieval word matrix and the keyword matrix includes: multiplying the retrieval word matrix by the keyword matrix to generate an intermediate matrix with one row and multiple columns, where one column of the intermediate matrix includes the retrieval score of a first keyword set; storing the intermediate matrix as the last row into the keyword matrix to generate the target matrix, where each column of the target matrix includes at least one first keyword set and the retrieval score of this first keyword set.

2. The retrieval method according to claim 1, characterized in that, before the step of constructing the target matrix according to the retrieval word matrix and the keyword matrix, the retrieval method further includes: obtaining attention keywords; removing a part of keywords from the keyword matrix according to the attention keywords to filter the keyword matrix; performing dimensionality reduction processing on the filtered keyword matrix using the singular value decomposition method.

3. The retrieval method according to claim 2, characterized in that, Determining a first retrieval coefficient for each retrieval term in a preset retrieval term dictionary according to the target term set includes: Removing a part of the retrieval terms from the retrieval term dictionary according to the concerned keyword to filter the retrieval term dictionary; For at least one retrieval term in the filtered retrieval term dictionary, Determining the first retrieval coefficient of the retrieval term according to whether there is a target term matching the retrieval term in the target term set.

4. The retrieval method according to claim 3, wherein, The retrieval term dictionary includes the weights of at least one retrieval term. Determining the first retrieval coefficient of the retrieval term according to whether there is a target term matching the retrieval term in the target term set includes: When there is a target term matching the retrieval term in the target term set, using the weight of the retrieval term as the first retrieval coefficient of the retrieval term; When there is no target term matching the retrieval term in the target term set, setting the first retrieval coefficient of the retrieval term to zero.

5. The retrieval method according to any one of claims 1 to 4, wherein, Generating a retrieval result according to the target matrix includes: Screening out at least a part from the first keyword sets in the target matrix according to the retrieval scores of each first keyword set in the target matrix; Obtaining the audio files to which the screened first keyword sets belong; Sorting the obtained audio files according to the retrieval scores to generate the retrieval result.

6. The retrieval method according to any one of claims 1 to 4, wherein, The retrieval method further includes: Obtaining newly added audio files; Extracting keywords from the newly added audio files according to the retrieval term dictionary to obtain a third keyword set of the newly added audio files; Updating the keyword matrix according to the third keyword set.

7. A retrieval device, wherein, comprises: A first acquisition module, configured to acquire at least one target term to obtain a target term set; A first retrieval coefficient determination module, configured to determine a first retrieval coefficient for each retrieval term in a preset retrieval term dictionary according to the target term set; A retrieval term matrix construction module, configured to construct a retrieval term matrix according to the retrieval term dictionary and the determined first retrieval coefficients; A second acquisition module, configured to acquire a keyword matrix, where the keyword matrix includes first keyword sets of at least one audio file; A target matrix construction module, configured to construct a target matrix according to the retrieval term matrix and the keyword matrix, where the target matrix includes at least one first keyword set and the retrieval scores of this first keyword set generated according to the retrieval term matrix; A retrieval result generation module, configured to generate a retrieval result according to the target matrix; The second acquisition module is further configured to construct the keyword matrix through the following steps: acquiring at least one audio file; extracting keywords from each audio file according to the retrieval term dictionary to obtain a first keyword set of each audio file; and forming the keyword matrix from the obtained first keyword sets; The second acquisition module is further configured to convert the audio file into a text file; extract keywords matching the search terms in the search term dictionary from the text file to obtain a second keyword set; determine a second search coefficient for each keyword in the second keyword set according to the number of occurrences of each keyword in the text file; and construct the first keyword set according to each keyword in the second keyword set and the second search coefficient of each keyword. The keyword matrix is a multi-row and multi-column matrix, which includes the first keyword sets of multiple audio files. One column of the keyword matrix includes one first keyword set, and one row of the keyword matrix includes the second search coefficients of multiple keywords with the same semantics. The search term matrix is a one-row and multi-column matrix, and one column of the search term matrix includes the first search coefficient of one search term. The target matrix construction module is further configured to multiply the search term matrix by the keyword matrix to generate a one-row and multi-column intermediate matrix, where one column of the intermediate matrix includes the search score of one first keyword set; and store the intermediate matrix as the last row into the keyword matrix to generate the target matrix, where each column of the target matrix includes at least one first keyword set and the search score of the first keyword set.

8. An electronic device Characterized in that Comprising: One or more processors; A storage device for storing one or more programs, Wherein, when the one or more programs are executed by the one or more processors, the one or more processors are caused to execute the retrieval method according to any one of claims 1 to 6.

9. A computer-readable storage medium Characterized in that An executable instruction is stored thereon, and when the instruction is executed by a processor, the processor is caused to execute the retrieval method according to any one of claims 1 to 6.

10. A computer program product Characterized in that Comprising a computer program, and when the computer program is executed by a processor, the retrieval method according to any one of claims 1 to 6 is implemented.

Images