A method, device and electronic device for locating playback data based on knowledge graph

Abstract

The present application provides a method, device and electronic device for locating playback data based on a knowledge graph. The playback data locating method includes: obtaining at least one first search keyword to be located; inputting each first search keyword into a constructed knowledge graph database, determining multiple candidate search playback data that match the first search keyword, and second search keywords in each candidate search playback data that have a semantic association with the first search keyword; based on the weight coefficient corresponding to each second search keyword, screening out target search playback data from multiple candidate search playback data, and determining the target moment when the second search keyword appears in the target search playback data. The present application can directly find the required target playback data among a large number of playback data and locate it to the desired playback position, thereby improving the efficiency and accuracy of the target playback data search and the positioning of the target moment in the target playback data.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, device and electronic device for locating playback data based on a knowledge graph. Background Art

[0002] With the development of electronic products, people have an increasing demand for data storage of electronic products. Therefore, the storage space for data in electronic products is getting larger and larger. However, as more and more playback data needs to be stored in electronic products, such as audio data, video data, etc., it is difficult to directly find the required target playback data and locate the desired playback position among the numerous stored playback data. Usually, the approximate position can only be manually determined based on the relevant memory of the operator, which takes a long time and results in low efficiency and accuracy in obtaining the target playback position of the target playback data. Summary of the invention

[0003] In view of this, the purpose of the present application is to provide a playback data positioning method, device and electronic device based on a knowledge graph, which can directly find the required target playback data among a large amount of playback data and locate it to the desired playback position, thereby improving the efficiency and accuracy of target playback data search and positioning of target moments in target playback data.

[0004] The embodiment of the present application provides a method for locating playback data based on a knowledge graph, and the method for locating playback data includes:

[0005] Obtain the search information to be located;

[0006] Performing word segmentation processing on the search information to determine at least one first search keyword in the search information;

[0007] Input each of the first search keywords into the constructed knowledge graph database, and determine a plurality of candidate search playback data that match the first search keyword, and a second search keyword that has a semantic association with the first search keyword in each of the candidate search playback data; wherein the knowledge graph database includes a playback data knowledge graph, a weight coefficient of each preset keyword in each preset search playback data, and a preset time at which each preset keyword appears in each preset search playback data;

[0008] Based on the weight coefficients corresponding to the respective second search keywords, target search and playback data are screened out from a plurality of candidate search and playback data, and a target time at which the second search keyword appears in the target search and playback data is determined.

[0009] Further, the weight coefficient of each preset keyword in each preset search and playback data is determined in the following manner;

[0010] Input each preset search and playback data into a feature extraction layer in a trained keyword scoring model to determine at least one preset keyword in the preset search and playback data;

[0011] Each of the preset keywords is input into the semantic similarity scoring layer in the trained keyword scoring model, and a weight coefficient of each of the preset keywords in the preset search and playback data is determined.

[0012] Furthermore, the trained keyword scoring model is determined by the following method:

[0013] Acquire a plurality of sample search and playback data and a sample label of each sample search and playback data; the sample label is used to characterize a real sample weight coefficient of each sample keyword in the corresponding sample search and playback data in each sample search and playback data;

[0014] Inputting the sample search playback data and the sample label into an initial keyword scoring model, training the initial keyword scoring model, and determining a preset sample weight coefficient of each sample keyword in each sample search playback data;

[0015] When the loss value between the preset sample weight coefficient and the real sample weight coefficient is less than a preset threshold, the training is terminated and the trained keyword scoring model is determined.

[0016] Furthermore, the step of inputting each of the first search keywords into the constructed knowledge graph database to determine a plurality of candidate search playback data matching the first search keyword includes:

[0017] Based on the first search keyword and the playback data knowledge graph in the knowledge graph database, a plurality of candidate search playback data having a semantic association with the first search keyword or a word frequency coupling degree greater than a preset value are determined.

[0018] Furthermore, the step of selecting target search and playback data from a plurality of candidate search and playback data based on the weight coefficients corresponding to the respective second search keywords includes:

[0019] Filtering out, from the plurality of second search keywords, a target search keyword whose corresponding weight coefficient exceeds a preset weight coefficient;

[0020] The candidate search playback data corresponding to the target search keyword is determined as the target search playback data.

[0021] The embodiment of the present application further provides a playback data locating device based on a knowledge graph, the playback data locating device comprising:

[0022] An acquisition module, used to acquire the search information to be located;

[0023] A first determination module, configured to perform word segmentation processing on the search information to determine at least one first search keyword in the search information;

[0024] A second determination module is configured to input each of the first search keywords into a constructed knowledge graph database, and determine a plurality of candidate search playback data that match the first search keyword, and a second search keyword that is semantically associated with the first search keyword in each of the candidate search playback data; wherein the knowledge graph database includes a playback data knowledge graph, a weight coefficient of each preset keyword in each preset search playback data, and a preset time at which each preset keyword appears in each of the preset search playback data;

[0025] The third determination module is used to filter out target search playback data from multiple candidate search playback data based on the weight coefficients corresponding to each of the second search keywords, and determine the target time when the second search keyword appears in the target search playback data.

[0026] Furthermore, the second determination module determines the weight coefficient of each preset keyword in each preset search and playback data in the following manner:

[0027] Input each preset search and playback data into a feature extraction layer in a trained keyword scoring model to determine at least one preset keyword in the preset search and playback data;

[0028] Each of the preset keywords is input into the semantic similarity scoring layer in the trained keyword scoring model, and a weight coefficient of each of the preset keywords in the preset search and playback data is determined.

[0029] Furthermore, the trained keyword scoring model is determined by the following method:

[0030] Acquire a plurality of sample search and playback data and a sample label of each sample search and playback data; the sample label is used to characterize a real sample weight coefficient of each sample keyword in the corresponding sample search and playback data in each sample search and playback data;

[0031] Inputting the sample search playback data and the sample label into an initial keyword scoring model, training the initial keyword scoring model, and determining a preset sample weight coefficient of each sample keyword in each sample search playback data;

[0032] When the loss value between the preset sample weight coefficient and the real sample weight coefficient is less than a preset threshold, the training is terminated and the trained keyword scoring model is determined.

[0033] An embodiment of the present application also provides an electronic device, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the playback data positioning method as described above are performed.

[0034] An embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the playback data positioning method as described above are executed.

[0035] The knowledge graph-based playback data positioning method, device and electronic device provided in the embodiments of the present application, compared with the prior art, the embodiments provided by the present application determine multiple candidate search playback data matching the first search keyword by inputting the first search keyword in the search information into a constructed knowledge graph database, and further determine the target search playback data from the multiple candidate search playback data based on the weight coefficient of the second search keyword that has a semantic association with the first search keyword, and determine the target moment when the second search keyword appears in the target search playback data. The required target playback data can be directly found in a large number of playback data and located to the desired playback position, thereby improving the efficiency and accuracy of the target playback data search and the positioning of the target moment in the target playback data.

[0036] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 One of the flow charts of a method for locating playback data based on a knowledge graph provided in an embodiment of the present application is shown;

[0039] Figure 2A second flowchart of a method for locating playback data based on a knowledge graph provided in an embodiment of the present application is shown;

[0040] Figure 3 A flow chart of an embodiment of a method for locating playback data based on a knowledge graph provided in an embodiment of the present application is provided for the present application;

[0041] Figure 4 A schematic diagram of the structure of a playback data locating device based on a knowledge graph provided in an embodiment of the present application is shown;

[0042] Figure 5 A schematic structural diagram of an electronic device provided in an embodiment of the present application is shown.

[0043] In the figure:

[0044] 400 - playback data positioning device; 410 - acquisition module; 420 - first determination module; 430 - second determination module; 440 - third determination module; 500 - electronic device; 510 - processor; 520 - memory; 530 - bus. DETAILED DESCRIPTION

[0045] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application usually described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, each other embodiment obtained by those skilled in the art without making creative work belongs to the scope of protection of the present application.

[0046] First, the application scenarios to which this application is applicable are introduced. Through research, it is found that in the prior art, as more and more playback data needs to be stored in electronic products, such as audio data, video data, etc., it is difficult to directly find the required target playback data and locate the desired playback position among the many stored playback data. Usually, the approximate position can only be manually determined based on the relevant memory of the operator, which takes a long time and results in low efficiency and accuracy in obtaining the target playback position of the target playback data.

[0047] Based on this, the embodiments of the present application provide a playback data positioning method, device and electronic device based on a knowledge graph, which can directly find the required target playback data among a large amount of playback data and locate the desired playback position, thereby improving the efficiency and accuracy of target playback data search and positioning of target moments in target playback data.

[0048] See also Figure 1 , Figure 1 This is one of the flowcharts of a method for locating playback data based on a knowledge graph provided in an embodiment of the present application. Figure 1 As shown in , the method for locating playback data based on knowledge graph provided in the embodiment of the present application includes the following steps:

[0049] S101: Obtain search information to be located.

[0050] In this step, the search information to be located can be characterized as a search sentence containing at least one first search keyword, and the search sentence is generally text information, and in special application scenarios, it can also be audio information or video information.

[0051] Here, when the search sentence is audio information or video information, the audio information or video information is converted into text through speech recognition technology, converted into corresponding text information, and the converted text information is determined as the search information to be located.

[0052] Among them, special application scenarios include but are not limited to the entity corresponding to the input search information being a recording device, a video file, and a person who is inconvenient to perform related operations.

[0053] In this way, the search information to be located can be input by relevant staff or by irrelevant staff who have no idea about the historical data corresponding to the search information. In other words, the users of the playback data locating method provided by the present application can be not only relevant staff who are clear about the operation background, but also other staff who have the need to locate playback data. This avoids the limitation of traditional search methods that only relevant staff who have an understanding of historical data can locate relevant playback data.

[0054] S102: Perform word segmentation processing on the search information to determine at least one first search keyword in the search information.

[0055] In this step, the search information is segmented and normalized according to the semantics and word frequency of the sentences in the text information, so as to determine at least one first search keyword in the search information.

[0056] Here, the search information is segmented according to the semantics corresponding to the sentences in the text information, which can be specifically:

[0057] The semantic information corresponding to the sentence of the text information where the search information is located and the preset standard vocabulary library are obtained.

[0058] The search information is semantically segmented according to a preset standard vocabulary library, spaces, the semantic information and punctuation marks to determine at least one first search keyword in the search information.

[0059] Alternatively, the search information may be segmented according to the word frequency corresponding to the sentence in the text information as follows:

[0060] Get the preset standard vocabulary library.

[0061] The search information is segmented according to a preset standard vocabulary library to determine the vocabulary in the search information.

[0062] The words in the search information are sorted and divided according to word frequency to determine at least one first search keyword in the search information.

[0063] S103. Input each of the first search keywords into the constructed knowledge graph database, determine multiple candidate search playback data that match the first search keyword, and second search keywords that are semantically associated with the first search keyword in each of the candidate search playback data; wherein the knowledge graph database includes a playback data knowledge graph, a weight coefficient of each preset keyword in each preset search playback data, and a preset time when each preset keyword appears in each of the preset search playback data.

[0064] In this step, since the constructed knowledge graph database not only stores the playback data knowledge graph, but also stores the weight coefficient of each preset keyword in each preset search playback data and the preset time when each preset keyword appears in each preset search playback data, after each first search keyword is input into the constructed knowledge graph database, each first search keyword first enters the playback data knowledge graph, and through the playback data knowledge graph, multiple candidate search playback data matching a search keyword are determined, and then the weight coefficient of each preset keyword in each preset search playback data and the preset time when each preset keyword appears in each preset search playback data are determined through the trained keyword scoring model.

[0065] Here, the preset time when each preset keyword appears in each preset search and playback data can be set as a label of each preset keyword.

[0066] The playback data knowledge graph in the knowledge graph database is constructed by performing data mining, information processing, knowledge measurement, semantic analysis, and feature extraction on single or multiple preset search playback data, and constructing the playback data knowledge graph through the entities, relationships, attributes, and relationship attributes obtained after processing and analysis, specifically:

[0067] Perform semantic analysis on a single or multiple preset search playback data to determine knowledge elements for constructing a knowledge graph of the playback data, wherein the knowledge elements include but are not limited to the distribution of preset key words in the single or multiple preset search playback data, the relationship between each high-frequency word, and the attribute analysis between each preset key word.

[0068] The knowledge elements are integrated so that the preset search and playback data from different playback sources can be integrated, disambiguated, processed, reasoned and verified, and updated under the same framework specification, so that the knowledge elements and the framework specification can achieve the mutual integration of data, information, methods and experience, realize the correlation between semantics, and determine the constructed playback data knowledge graph.

[0069] In the above, the embodiment provided by the present application constructs a good playback data knowledge graph to help operators find more accurate candidate search playback data, make more comprehensive data screening and provide more accurate playback data-related information. Compared with the prior art, the text and time information converted from the entire playback data are stored. The embodiment provided by the present application only needs to store the preset information of the preset keywords extracted from the constructed playback data knowledge graph, which saves a lot of memory resources while also improving operation efficiency and ensuring the accuracy of the operation.

[0070] Furthermore, in step S103, the weight coefficient of each preset keyword in each preset search and playback data is determined through the following sub-steps:

[0071] Sub-step 1031: input each preset search and playback data into the feature extraction layer of the trained keyword scoring model to determine at least one preset keyword in the preset search and playback data.

[0072] Here, firstly, the data features of each preset search and playback data are extracted, and at least one preset keyword in each preset search and playback data is determined.

[0073] Among them, the basis provided by the feature extraction layer can be designed and combined according to different actual application scenarios. For example, in the embodiment provided in the present application, the feature extraction of the feature extraction layer can be based on but not limited to the semantics and word frequency of each preset search and playback data.

[0074] Sub-step 1032: input each of the preset keywords into the semantic similarity scoring layer in the trained keyword scoring model, and determine the weight coefficient of each of the preset keywords in the preset search and playback data.

[0075] Here, the trained keyword scoring model is determined in the following way:

[0076] A plurality of sample search and playback data and a sample label of each sample search and playback data are obtained; the sample label is used to characterize a real sample weight coefficient of each sample keyword in the corresponding sample search and playback data in each sample search and playback data.

[0077] The sample search and playback data and the sample labels are input into an initial keyword scoring model, the initial keyword scoring model is trained, and a preset sample weight coefficient of each sample keyword in each sample search and playback data is determined.

[0078] When the loss value between the preset sample weight coefficient and the real sample weight coefficient is less than a preset threshold, the training is terminated and the trained keyword scoring model is determined.

[0079] In the above, after determining the weight coefficient of each preset keyword in the preset search and playback data, the weight coefficient of each preset keyword in each preset search and playback data and the preset time when each preset keyword appears in each preset search and playback data are all stored in the knowledge graph database.

[0080] Here, the preset time is the time information when each preset keyword appears in the corresponding preset search and playback data.

[0081] Furthermore, in step S103, each of the first search keywords is input into the constructed knowledge graph database to determine a plurality of candidate search playback data matching the first search keyword, including:

[0082] Based on the first search keyword and the playback data knowledge graph in the knowledge graph database, a plurality of candidate search playback data having a semantic association with the first search keyword or a word frequency coupling degree greater than a preset value are determined.

[0083] Here, before determining multiple candidate search playback data, a coupling degree preset value is first set. At this time, when the first search keyword is input into the playback data knowledge graph in the knowledge graph database, search playback data that has a semantic association or word frequency coupling degree with the first search keyword is found, and then the real semantic association value and word frequency coupling value between the search playback data and the first search keyword are determined by using the weighted average method of the semantic association or word frequency coupling degree or other custom calculation methods. When the above-mentioned real semantic association value and word frequency coupling value are greater than the preset value, the search playback data greater than the preset value is determined as multiple candidate search playback data that match the first search keyword.

[0084] The candidate search and playback data may be one or more, and the word frequency coupling value between the search and playback data and the first search keyword is determined based on the word repetition between the search and playback data and the first search keyword, specifically:

[0085] The number of words in the search playback data that are semantically associated with the first search keyword is determined.

[0086] The ratio between the number of the above words and the first search keyword is determined as the word frequency coupling value between the search playback data and the first search keyword.

[0087] In the above, the clustering of multiple candidate search playback data that are semantically associated with the first search keyword can quantify the semantic connection between semantic nodes through semantic symbiosis.

[0088] Among them, the higher the semantic symbiosis between two semantic nodes, the greater the probability that the two first search keywords appear in the same playback data.

[0089] S104: Filter target search and playback data from a plurality of candidate search and playback data based on the weight coefficients corresponding to the respective second search keywords, and determine a target time at which the second search keyword appears in the target search and playback data.

[0090] In this step, the second search keyword is a search keyword in the preset keywords that has a semantic association with the first search keyword. Here, the weight coefficient of the second search keyword is stored in the constructed knowledge graph database. After determining multiple candidate search playback data, the multiple candidate search playback data will be directly displayed in sequence according to the weight coefficient corresponding to the second search keyword. The staff can directly view the weight coefficient corresponding to each second search keyword in the candidate list composed of multiple candidate search playback data. At this time, the staff can select the second search keyword of one or any candidate search playback data according to actual needs. At this time, the selected candidate search playback data is the target search playback data, and the weight coefficient corresponding to the second search keyword in the target search playback data must be greater than the preset weight coefficient. At this time, the staff can directly operate the second search keyword through any operation instruction. At this time, the target search playback data can be directly located at the target time when the second search keyword appears.

[0091] Here, after determining the target time when the second search keyword appears in the target search playback data, the target search playback data related to the second search keyword at the target time will be displayed to the staff. Here, the content displayed to the staff is described with the following specific embodiments: for example, the audio and video of the meeting minutes in the preset search playback data that are semantically related to the second search keyword can be displayed; or, the educational learning audio and video materials in the preset search playback data that are semantically related to the second search keyword can be displayed.

[0092] Compared with the methods in the prior art, the playback data positioning method provided in the embodiments of the present application inputs the first search keyword in the search information into a constructed knowledge graph database, determines multiple candidate search playback data matching the first search keyword, and further determines the target search playback data from the multiple candidate search playback data based on the weight coefficient of the second search keyword that is semantically associated with the first search keyword, and determines the target moment when the second search keyword appears in the target search playback data. The required target playback data can be directly found among a large number of playback data and located to the desired playback position, thereby improving the efficiency and accuracy of the target playback data search and the positioning of the target moment in the target playback data.

[0093] See also Figure 2 , Figure 2 This is the second flowchart of a method for locating playback data based on a knowledge graph in this application. Figure 2 As shown in , the method for locating playback data based on knowledge graph provided in the embodiment of the present application includes:

[0094] S201: Obtain search information to be located.

[0095] S202: Perform word segmentation processing on the search information to determine at least one first search keyword in the search information.

[0096] S203. Input each of the first search keywords into the constructed knowledge graph database, determine multiple candidate search playback data that match the first search keyword, and second search keywords that are semantically associated with the first search keyword in each of the candidate search playback data; wherein the knowledge graph database includes a playback data knowledge graph, a weight coefficient of each preset keyword in each preset search playback data, and a preset time when each preset keyword appears in each of the preset search playback data.

[0097] S204: Filter out target search keywords whose corresponding weight coefficients exceed preset weight coefficients from the plurality of second search keywords.

[0098] In this step, at least one second search keyword whose weight coefficient exceeds the preset weight coefficient is the target search keyword.

[0099] Here, the setting of the preset weight coefficient can be customized according to the actual needs of different application scenarios.

[0100] Among them, the preset weight coefficient processing of the second search keyword is set according to the conventional semantic relevance and word frequency coupling. The weighted weight coefficient can also be set in combination with the time duration and time position information in the search playback data corresponding to the second search keyword, and finally the weight coefficient of the second search keyword is determined.

[0101] S205: Determine the candidate search playback data corresponding to the target search keyword as the target search playback data, and determine the target time when the second search keyword appears in the target search playback data.

[0102] Here, after determining the target search playback data, the operator can select a second search keyword according to the actual application scenario, and directly locate the target moment that appears in the target search playback data by selecting the second search keyword.

[0103] Among them, the description of S201 to S203 can refer to the description of S101 to S103, and can achieve the same technical effect, which is not repeated here.

[0104] Here, the specific process of the playback data positioning method provided by the present application is described through an embodiment. Figure 3 As shown, Figure 3A flowchart of an embodiment of a method for locating playback data based on a knowledge graph provided in an embodiment of the present application is provided for the present application, wherein the embodiment includes the following steps:

[0105] S301. Extract, mine, process and measure knowledge of audio and video features in preset search and playback data to determine initial knowledge graph elements.

[0106] S302. Input the preset search and playback data into the preset knowledge graph framework platform for semantic analysis to determine the distribution of each preset keyword, the relationship between each high-frequency preset keyword in each preset keyword, and the attribute analysis between each preset keyword, and update the initial knowledge graph elements according to the above analysis results to determine the target knowledge graph elements.

[0107] S303: Input the target knowledge graph elements into the preset knowledge graph framework platform for knowledge fusion to determine the playback data knowledge graph.

[0108] Among them, the target knowledge graph elements are input into the preset knowledge graph framework platform for knowledge fusion, so that the playback data from different sources can be integrated, disambiguated, processed, reasoned and verified, and updated under the same framework specifications, so that the playback data can achieve the integration of data, information, methods, experience and ideas, and use the semantic correlation between preset keywords to form a playback data knowledge graph.

[0109] S304: Based on the trained keyword scoring model, determine the weight coefficient of each preset keyword in the playback data knowledge graph, and record the preset time when each preset keyword appears in each preset search playback data.

[0110] S305. Store the preset time at which each preset keyword appears in each preset search and playback data, the weight coefficient of each preset keyword, and the playback data knowledge graph in the constructed knowledge graph database.

[0111] S306: Perform word segmentation and normalization processing on the search information to be located input by the staff, and determine at least one first search keyword in the search information.

[0112] S307: Input each first search keyword into the constructed knowledge graph database, determine multiple candidate search playback data that match the first search keyword, and second search keywords that are semantically associated with the first search keyword in each candidate search playback data.

[0113] S308. Sort the second search keyword according to the weight coefficient, display the related candidate search playbacks, filter out the target search playback data from the multiple candidate search playback data, determine the target moment when the second search keyword appears in the target search playback data, and locate the target search playback data to the target moment.

[0114] Compared with the methods in the prior art, the playback data positioning method provided in the embodiments of the present application inputs the first search keyword in the search information into a constructed knowledge graph database, determines multiple candidate search playback data matching the first search keyword, and further determines the target search playback data from the multiple candidate search playback data based on the weight coefficient of the second search keyword that is semantically associated with the first search keyword, and determines the target moment when the second search keyword appears in the target search playback data. The required target playback data can be directly found among a large number of playback data and located to the desired playback position, thereby improving the efficiency and accuracy of the target playback data search and the positioning of the target moment in the target playback data.

[0115] See also Figure 4 , Figure 4 This is a schematic diagram of the structure of a playback data positioning device based on a knowledge graph provided in an embodiment of the present application. Figure 4 As shown in , the playback data positioning device 400 includes:

[0116] The acquisition module 410 is used to acquire the search information to be located.

[0117] The first determination module 420 is configured to perform word segmentation processing on the search information to determine at least one first search keyword in the search information.

[0118] The second determination module 430 is used to input each of the first search keywords into the constructed knowledge graph database, determine multiple candidate search playback data matching the first search keyword, and second search keywords in each of the candidate search playback data that are semantically associated with the first search keyword; wherein the knowledge graph database includes a playback data knowledge graph, a weight coefficient of each preset keyword in each preset search playback data, and a preset time at which each preset keyword appears in each of the preset search playback data.

[0119] The third determination module 440 is used to filter out target search playback data from multiple candidate search playback data based on the weight coefficients corresponding to each of the second search keywords, and determine the target time when the second search keyword appears in the target search playback data.

[0120] Furthermore, the second determination module 430 determines the weight coefficient of each preset keyword in each preset search and playback data in the following manner.

[0121] Each preset search and playback data is input into a feature extraction layer in a trained keyword scoring model to determine at least one preset keyword in the preset search and playback data.

[0122] Each of the preset keywords is input into the semantic similarity scoring layer in the trained keyword scoring model, and a weight coefficient of each of the preset keywords in the preset search and playback data is determined.

[0123] Furthermore, the trained keyword scoring model is determined in the following manner.

[0124] A plurality of sample search and playback data and a sample label of each sample search and playback data are obtained; the sample label is used to characterize a real sample weight coefficient of each sample keyword in the corresponding sample search and playback data in each sample search and playback data.

[0125] The sample search and playback data and the sample labels are input into an initial keyword scoring model, the initial keyword scoring model is trained, and a preset sample weight coefficient of each sample keyword in each sample search and playback data is determined.

[0126] When the loss value between the preset sample weight coefficient and the real sample weight coefficient is less than a preset threshold, the training is terminated and the trained keyword scoring model is determined.

[0127] Compared with the prior art, the playback data positioning device 400 provided in the embodiment of the present application inputs the first search keyword in the search information into a constructed knowledge graph database, determines multiple candidate search playback data matching the first search keyword, and further determines the target search playback data from the multiple candidate search playback data based on the weight coefficient of the second search keyword that is semantically associated with the first search keyword, and determines the target moment when the second search keyword appears in the target search playback data. The required target playback data can be directly found in a large number of playback data and located at the desired playback position, thereby improving the efficiency and accuracy of target playback data search and positioning of the target moment in the target playback data.

[0128] See also Figure 5 , Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 5 As shown in , the electronic device 500 includes a processor 510 , a memory 520 and a bus 530 .

[0129] The memory 520 stores machine-readable instructions executable by the processor 510. When the electronic device 500 is running, the processor 510 communicates with the memory 520 via the bus 530. When the machine-readable instructions are executed by the processor 510, the above-mentioned Figure 1 as well as Figure 2 The steps of the playback data positioning method in the method embodiment shown, the specific implementation method can be found in the method embodiment, and will not be repeated here.

[0130] The present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the computer program can execute the above-mentioned Figure 1 as well as Figure 2 The steps of the playback data positioning method in the method embodiment shown, the specific implementation method can be found in the method embodiment, and will not be repeated here.

[0131] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0132] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0133] 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 distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0134] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0135] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application can essentially be embodied in the form of a software product, or in other words, the part that contributes to the prior art or the part of the technical solution. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0136] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed in the present application, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A playback data location method based on knowledge graph, It is characterized in that The playback data positioning method comprises: Obtain the search information to be located; Performing word segmentation processing on the search information to determine at least one first search keyword in the search information; Input each of the first search keywords into the constructed knowledge graph database, and determine a plurality of candidate search playback data that match the first search keyword, and a second search keyword that has a semantic association with the first search keyword in each of the candidate search playback data; wherein the knowledge graph database includes a playback data knowledge graph, a weight coefficient of each preset keyword in each preset search playback data, and a preset time at which each preset keyword appears in each preset search playback data; The step of inputting each of the first search keywords into the constructed knowledge graph database to determine a plurality of candidate search playback data that matches the first search keyword includes: based on the first search keyword and the playback data knowledge graph in the knowledge graph database, determining a plurality of candidate search playback data that have a semantic association with the first search keyword or a word frequency coupling degree greater than a preset value; Based on the weight coefficients corresponding to the respective second search keywords, target search and playback data are selected from a plurality of candidate search and playback data, and a target time at which the second search keyword appears in the target search and playback data is determined; The method of screening out target search playback data from a plurality of candidate search playback data based on the weight coefficient corresponding to each of the second search keywords comprises: screening out a target search keyword whose corresponding weight coefficient exceeds a preset weight coefficient from a plurality of second search keywords; and determining the candidate search playback data corresponding to the target search keyword as the target search playback data.

2. The playback data positioning method according to claim 1, It is characterized in that Determine the weight coefficient of each preset keyword in each preset search and playback data in the following manner; Input each preset search and playback data into a feature extraction layer in a trained keyword scoring model to determine at least one preset keyword in the preset search and playback data; Each of the preset keywords is input into the semantic similarity scoring layer in the trained keyword scoring model, and a weight coefficient of each of the preset keywords in the preset search and playback data is determined.

3. The playback data positioning method according to claim 2, It is characterized in that Determine the trained keyword scoring model in the following ways: Acquire a plurality of sample search and playback data and a sample label of each sample search and playback data; the sample label is used to characterize a real sample weight coefficient of each sample keyword in the corresponding sample search and playback data in each sample search and playback data; Inputting the sample search playback data and the sample label into an initial keyword scoring model, training the initial keyword scoring model, and determining a preset sample weight coefficient of each sample keyword in each sample search playback data; When the loss value between the preset sample weight coefficient and the real sample weight coefficient is less than a preset threshold, the training is terminated and the trained keyword scoring model is determined.

4. A playback data positioning device based on knowledge graph, It is characterized in that The playback data locating device comprises: An acquisition module, used to acquire the search information to be located; A first determination module, configured to perform word segmentation processing on the search information to determine at least one first search keyword in the search information; A second determination module is configured to input each of the first search keywords into a constructed knowledge graph database, and determine a plurality of candidate search playback data that match the first search keyword, and a second search keyword that is semantically associated with the first search keyword in each of the candidate search playback data; wherein the knowledge graph database includes a playback data knowledge graph, a weight coefficient of each preset keyword in each preset search playback data, and a preset time at which each preset keyword appears in each of the preset search playback data; The second determination module is specifically configured to determine, based on the first search keyword and the playback data knowledge graph in the knowledge graph database, a plurality of candidate search playback data that have a semantic association with the first search keyword or a word frequency coupling degree greater than a preset value; A third determination module is used to screen target search playback data from a plurality of candidate search playback data based on the weight coefficients corresponding to each of the second search keywords, and determine a target time when the second search keyword appears in the target search playback data; The third determination module is specifically used to screen out a target search keyword whose corresponding weight coefficient exceeds a preset weight coefficient from multiple second search keywords; and determine the candidate search playback data corresponding to the target search keyword as the target search playback data.

5. The playback data positioning device according to claim 4, It is characterized in that The second determination module specifically determines the weight coefficient of each preset keyword in each preset search and playback data in the following manner; Input each preset search and playback data into a feature extraction layer in a trained keyword scoring model to determine at least one preset keyword in the preset search and playback data; Each of the preset keywords is input into the semantic similarity scoring layer in the trained keyword scoring model, and a weight coefficient of each of the preset keywords in the preset search and playback data is determined.

6. The playback data positioning device according to claim 5, It is characterized in that Determine the trained keyword scoring model in the following ways: Acquire a plurality of sample search and playback data and a sample label of each sample search and playback data; the sample label is used to characterize a real sample weight coefficient of each sample keyword in the corresponding sample search and playback data in each sample search and playback data; Inputting the sample search playback data and the sample label into an initial keyword scoring model, training the initial keyword scoring model, and determining a preset sample weight coefficient of each sample keyword in each sample search playback data; When the loss value between the preset sample weight coefficient and the real sample weight coefficient is less than a preset threshold, the training is terminated and the trained keyword scoring model is determined.

7. An electronic device, It is characterized in that include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate via the bus, and the machine-readable instructions are executed by the processor to perform the steps of the playback data positioning method as described in any one of claims 1 to 3 above.

8. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the playback data positioning method as described in any one of claims 1 to 3 are executed.

Images