Data retrieval method and device based on artificial intelligence

Abstract

The invention provides a data retrieval method and device based on artificial intelligence, and relates to the technical field of computers, in particular to the technical field of artificial intelligence. According to the specific implementation scheme, in response to a received data retrieval request, an intention parameter representing a retrieval intention of a user is obtained according to the data retrieval request; determining a target information channel matched with the intention parameter in each candidate information channel; according to the data retrieval request, obtaining channel input information input to a target information channel for data retrieval; obtaining a channel retrieval result obtained by inputting the channel input information into the target information channel for data retrieval; and performing data fusion on the obtained retrieval results of each channel, and feeding back the obtained fusion result as a data retrieval result. By applying the data retrieval scheme provided by the embodiment of the invention, the user experience can be improved.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technologies, and in particular to the field of artificial intelligence technologies, and particularly to a data retrieval method and apparatus based on artificial intelligence. Background Art

[0002] Nowadays, most data platforms have a data retrieval function, and users can perform data retrieval in the data platform to obtain retrieval results.

[0003] A data platform may provide multiple information channels. During the data retrieval process, a user may initiate a data retrieval request carrying data with a specified information channel in the data platform. In this way, in response to receiving the data retrieval request, the data platform can perform a retrieval in the specified information channel, obtain a retrieval result, and feedback the retrieval result to the user. Summary of the Invention

[0004] The present disclosure provides a data retrieval method and apparatus based on artificial intelligence.

[0005] In a first aspect, an embodiment of the present disclosure provides a data retrieval method based on artificial intelligence, including:

[0006] In response to receiving a data retrieval request, obtaining an intent parameter characterizing the retrieval intent of the user according to the data retrieval request;

[0007] Determining a target information channel that matches the intent parameter in each candidate information channel;

[0008] Obtaining channel input information input to the target information channel for data retrieval according to the data retrieval request;

[0009] Obtaining a channel retrieval result of inputting the channel input information to the target information channel for data retrieval;

[0010] Performing data fusion on the obtained channel retrieval results, and feeding back the obtained fusion result as the data retrieval result.

[0011] In a second aspect, an embodiment of the present disclosure provides a data retrieval apparatus based on artificial intelligence, including:

[0012] An intent acquisition module, configured to obtain an intent parameter characterizing the retrieval intent of the user according to the data retrieval request in response to receiving the data retrieval request;

[0013] A channel determination module, configured to determine a target information channel that matches the intent parameter in each candidate information channel;

[0014] An information acquisition module, configured to obtain channel input information that is input to the target information channel for data retrieval according to the data retrieval request;

[0015] A result acquisition module, configured to obtain a channel retrieval result of inputting the channel input information to the target information channel for data retrieval;

[0016] A result feedback module, configured to perform data fusion on the obtained channel retrieval results, and use the obtained fusion result as the data retrieval result for feedback.

[0017] In a third aspect, an embodiment of the present disclosure provides an intelligent agent of artificial intelligence, configured to execute the method described in the foregoing first aspect.

[0018] In a fourth aspect, an embodiment of the present disclosure provides an electronic device, including:

[0019] At least one processor; and

[0020] A memory communicatively connected to the at least one processor; wherein,

[0021] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method described in the foregoing first aspect.

[0022] In a fifth aspect, an embodiment of the present disclosure provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the method described in the foregoing first aspect.

[0023] In a sixth aspect, an embodiment of the present disclosure provides a computer program product, including a computer program, and the computer program realizes the method described in the foregoing first aspect when executed by a processor.

[0024] As can be seen from the above, when retrieving data by applying the solution provided by the embodiment of the present disclosure, the data platform can accurately determine a target information channel that matches the intent parameter representing the user's retrieval intent according to the intent parameter, so that without the user specifying the information channel, the information channel that meets the user's intent can be determined, and thus data retrieval can be performed in the target information channel to obtain a channel retrieval result that meets the user's intent. By fusing the channel retrieval results and using the obtained fusion result as the data retrieval result for feedback, data that meets the user's intent can be fed back to the user. Therefore, by applying the data retrieval solution provided by the embodiment of the present disclosure, the accuracy of data retrieval can be improved and the user experience can be enhanced.

[0025] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. Brief Description of the Drawings

[0026] The drawings are used to better understand the present solution and do not constitute a limitation to the present disclosure. Among them:

[0027] Figure 1 is a schematic flowchart of the first data retrieval method provided by an embodiment of the present disclosure;

[0028] Figure 2 is a schematic flowchart of the second data retrieval method provided by an embodiment of the present disclosure;

[0029] Figure 3 is a schematic structural diagram of the first data retrieval device provided by an embodiment of the present disclosure;

[0030] Figure 4 is a block diagram of an electronic device for implementing the data retrieval method according to an embodiment of the present disclosure. Detailed Embodiments

[0031] The following describes exemplary embodiments of the present disclosure with reference to the accompanying drawings. Various details of the embodiments of the present disclosure are included to assist understanding, and they should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, descriptions of well-known functions and structures are omitted below for clarity and conciseness.

[0032] First, the application scenarios of the data retrieval solution provided by the embodiments of the present disclosure are described.

[0033] When a user is writing documents such as papers and reports, the user may need to consult materials related to the theme of the written document. In such a case, the user can initiate a data retrieval request in the data platform. In response to receiving the data retrieval request, the data platform can apply the data retrieval solution provided by the embodiments of the present disclosure to perform data retrieval, obtain the retrieval results, and feedback the retrieval results to the user. After the user consults the retrieval results, the user can write the document according to the information included in the retrieval results.

[0034] In addition, in addition to the scenario of retrieving materials related to the document written by the user as described above, the data retrieval solution provided by the embodiments of the present disclosure can also be applied to scenarios of retrieving other information, and the embodiments of the present disclosure do not limit this.

[0035] The following details the artificial intelligence-based data retrieval method and device provided by the embodiments of the present disclosure through specific embodiments.

[0036] See Figure 1 , Figure 1 which is a schematic flowchart of the first artificial intelligence-based data retrieval method provided by the embodiments of the present disclosure. In this embodiment, the above method includes the following steps S101-S105.

[0037] Step S101: In response to receiving a data retrieval request, obtain an intent parameter characterizing the user's retrieval intent according to the data retrieval request.

[0038] First, the above intent parameter is described.

[0039] The above intent parameter is used to characterize the user's retrieval intent. The user's retrieval intent can be understood as the content that the user wants to retrieve.

[0040] For example, if the user enters a keyword: "The Sound of the Wind", the data related to "The Sound of the Wind" may include TV dramas, music, news, or articles related to the sound of the wind, etc. For these data, what the user may want to retrieve is the music related to the sound of the wind. Therefore, the user's retrieval intent is that the user wants to retrieve the music related to the sound of the wind, and the above intent parameter is a parameter characterizing the meaning that the user wants to retrieve the music related to the sound of the wind.

[0041] Specifically, after receiving the data retrieval request, the platform can parse the parameters carried in the data retrieval request to obtain a parsing result, and obtain the above intent parameter according to the parsing result.

[0042] The parameters carried in the data retrieval request may belong to structured data or unstructured data. For example, the data retrieval request may carry parameters such as dates, names, locations, event keywords, etc. in a preset input format. Another example is that the data retrieval request may carry a piece of text, a string of characters, etc. input by the user. When the data platform parses the parameters carried in the data retrieval request, it can detect whether the parameters carried in the data retrieval request belong to structured data. If the parameters belong to structured data, the various parameters carried in the data retrieval request are determined according to the data parsing rules specified based on the preset input format as the parsing result; if the parameters belong to unstructured data, a parameter recognition algorithm, model, etc. can be used to recognize the parameters carried in the data retrieval request to obtain a recognition result as the parsing result.

[0043] For example, if the data retrieval request carries a piece of text input by the user, the data platform can perform text understanding on the text to obtain a parsing result representing the text semantics of the text.

[0044] After parsing to obtain the parsing result, the data platform can obtain an intent parameter characterizing the user's retrieval intent according to the parsing result.

[0045] In one embodiment of the present disclosure, the data platform may detect whether an intent parameter representing the user's retrieval intent is carried in the data retrieval request. If not, intent recognition is performed based on the data retrieval request to obtain the intent parameter. If it exists, the intent parameter carried in the data retrieval request is directly obtained.

[0046] After the platform parses the parameters carried in the data retrieval request, various parameters carried in the data retrieval request can be determined. In this case, the determined various parameters can be traversed to detect whether there is an intent parameter among the various parameters carried in the data retrieval request. If not, intent recognition is performed based on the parsing result to obtain the intent parameter. If it exists, the detected intent parameter is directly obtained.

[0047] By adopting this method, the data platform can quickly obtain the intent parameter when the data retrieval request contains the intent parameter, so as to perform data retrieval based on the intent parameter, which can improve the efficiency of data retrieval.

[0048] For the specific implementation manner of performing intent recognition based on the data retrieval request, reference can be made to the subsequent embodiments, which will not be elaborated here for the time being.

[0049] Step S102: Determine the target information channel that matches the intent parameter among the candidate information channels.

[0050] Specifically, the data in different candidate information channels may be different. For example, some candidate information channels may contain relatively rich news data, some candidate information channels may contain relatively rich film and television data, and some candidate information channels may contain relatively rich encyclopedia knowledge data; or for another example, the data in some candidate information channels may mostly be data with strong popular science but weak professionalism, and the data in some candidate information channels may mostly be data with strong professionalism. In this way, after the data platform determines the intent parameter, it can determine, among the candidate information channels, the information channel that can provide data that meets the user's intent as the target information channel according to the user intent represented by the intent parameter.

[0051] For example, if a user expects to write a student paper about tigers, the user can enter a piece of text in the data platform: Query information about tigers, generate a data retrieval request based on this text. The data platform can, in response to receiving this data retrieval request, parse the parameters carried in the data retrieval request, and according to the parsing result, obtain an intention parameter characterizing that the user expects to query articles related to tigers. Thus, based on this intention parameter, among the candidate information channels, an information channel that can provide data with relatively strong popular science nature but not high professionalism can be selected as the target information channel, or an information channel containing relatively rich data of encyclopedia knowledge type can be selected as the target information channel.

[0052] Step S103: Obtain the channel input information input to the target information channel for data retrieval according to the data retrieval request.

[0053] There may be one or more of the above-mentioned target information channels. For each target information channel, according to the data retrieval request, the channel input information input to this target information channel for data retrieval can be obtained.

[0054] For the specific implementation method of obtaining the channel input information according to the data retrieval request, reference can be made to the subsequent embodiments, which will not be elaborated here for the time being.

[0055] Step S104: Obtain the channel retrieval result of inputting the channel input information into the target information channel for data retrieval.

[0056] Specifically, for each target information channel, after obtaining the channel input information corresponding to this target information channel, the channel input information corresponding to this target information channel can be input into the target information channel for data retrieval to obtain the channel retrieval result feedback by this target information channel. In this way, when there are multiple target information channels, multiple channel retrieval results feedback by multiple target information channels can be obtained.

[0057] Step S105: Perform data fusion on the obtained channel retrieval results, and use the obtained fusion result as the data retrieval result for feedback.

[0058] Specifically, any one of the following two implementation methods can be used for data fusion to obtain the data fusion result.

[0059] In the first implementation method, the obtained channel retrieval results can be input into the trained data fusion model to obtain the data fusion result output by the data fusion model for data fusion of the channel retrieval results.

[0060] The above data fusion model may belong to an LLM (Large Language Model). The above data fusion model is used to perform data fusion on data. After the data platform obtains the retrieval results of each channel, it inputs the retrieval results of each channel into the data fusion model. The data fusion model can accurately perform data fusion on the retrieval results of each channel, output a relatively accurate data fusion result, and then use the data fusion result as the data retrieval result for feedback, which can improve the accuracy of data retrieval.

[0061] In the second implementation manner, some data fusion algorithms can be used to perform data fusion on the retrieval results of each channel.

[0062] For example, duplicate data removal can be performed on the retrieval results of each channel, that is, the duplicate data in the retrieval results of each channel is removed to obtain a data fusion result containing all the remaining data.

[0063] In addition to the duplicate data removal operation, operations such as integration and classification can also be performed on the remaining data.

[0064] Integration means that for data belonging to the same event in the retrieval results of each channel, these data are re-expressed to obtain the integrated data. For example, in the retrieval result of one channel, the occurrence time of event A is T1, and in the retrieval result of another channel, the occurrence location of event A is L1. In this way, during the data fusion process, these two channel retrieval results can be integrated to obtain the integrated data: the occurrence time of event A is T1 and the occurrence location is L1.

[0065] Classification is data classification. The data classification types can include pictures, videos, texts, etc. The data platform can classify the remaining data according to various data classification types.

[0066] The above duplicate data removal and integration operations can adopt efficient filtering and integration algorithms, which can ensure the high quality, high value, and high utilization rate of the output data.

[0067] As can be seen from the above, when retrieving data using the solution provided in the embodiments of the present disclosure, the data platform can accurately determine the target information channel that matches the intent parameter according to the intent parameter representing the user's retrieval intent. In this way, without the user specifying the information channel, the information channel that meets the user's intent can be determined, and thus data retrieval can be performed in the target information channel to obtain the channel retrieval result that meets the user's intent. By fusing the retrieval results of each channel and using the obtained fusion result as the data retrieval result for feedback, data that meets the user's intent can be fed back to the user. Therefore, applying the data retrieval solution provided in the embodiments of the present disclosure can improve the accuracy of data retrieval and enhance the user's usage experience.

[0068] Before obtaining the intention parameter representing the user's search intention based on the data retrieval request, the data platform may process the data retrieval request in the manner mentioned in step S201 in the following embodiment.

[0069] In one embodiment of the present disclosure, see Figure 2 , a flow chart of a second data retrieval method based on artificial intelligence is provided. In this embodiment, the method includes the following steps S201-S206.

[0070] Step S201: In response to receiving a data retrieval request, perform input parameter verification on the data retrieval request. If the verification passes, execute step S202.

[0071] Specifically, the data platform can use input parameter verification technology to perform input parameter verification on the data retrieval request. If the verification fails, for example, it is detected that the data retrieval request does not conform to the preset request format, or garbled characters are detected in the request, etc., there is no need to proceed to the next step. At this time, the user can be directly notified that the retrieval failed or the data retrieval request did not pass the input parameter verification, etc. If the verification passes, step S202 is executed.

[0072] Step S202: According to the data retrieval request, an intention parameter representing the user's retrieval intention is obtained.

[0073] Step S203: Determine a target information channel that matches the intention parameter among the candidate information channels.

[0074] Step S204: According to the data search request, channel input information input into the target information channel for data search is obtained.

[0075] Step S205: obtaining a channel search result of inputting the channel input information into the target information channel for data search.

[0076] Step S206: performing data fusion on the retrieval results obtained from each channel, and feeding back the obtained fusion result as the data retrieval result.

[0077] The above steps S202-S206 are the same as the above steps S101-S105, and will not be repeated here.

[0078] From the above, it can be seen that when the solution provided by the embodiment of the present disclosure is applied to retrieve data, the data retrieval request is checked for input parameters, and subsequent operations are performed only when the check passes. This can reduce the workload of the data platform and improve the accuracy of data retrieval.

[0079] The specific implementation method of performing intent recognition based on the data retrieval request mentioned in the above step S101 is described below.

[0080] In the first acquisition method, the data platform inputs the parameters carried in the data retrieval request into the trained intent recognition model, and obtains the intent recognition result output by the intent recognition model as the intent parameter.

[0081] The above-mentioned intent recognition model can belong to the LLM. The above-mentioned intent recognition model is used for intent recognition. After the data platform inputs the parameters carried in the data retrieval request into the intent recognition model, the intent recognition model can perform intent recognition based on the parameters carried in the data retrieval request to obtain the intent recognition result. In this way, after the data platform obtains the intent recognition result output by the intent recognition model, it can use this intent recognition result as the intent parameter.

[0082] The above-mentioned intent recognition model is a trained model with intent recognition ability. Using the intent recognition model for intent recognition can accurately obtain the intent parameter, so as to perform data retrieval based on the intent parameter, which can improve the accuracy of data retrieval.

[0083] In the second implementation method, the data platform can perform intent recognition based on the parameters carried in the data retrieval request by using intent recognition algorithms, functions, etc., and use the intent recognition result as the intent parameter.

[0084] The implementation method for determining the target information channel will be described below.

[0085] In an embodiment of the present disclosure, when determining the target information channel, the data platform can input the intent parameter into the trained channel determination model, obtain the channel determination results of each candidate information channel output by the channel determination model, and determine the target information channel among each candidate information channel according to the channel determination results.

[0086] Among them, the above-mentioned channel determination model can belong to the LLM. The above-mentioned channel determination model is used to determine whether to use each candidate information channel for data retrieval during the data retrieval process.

[0087] The channel determination result is the result determined by the channel determination model based on the intent parameter, which represents whether to use the candidate information channel for data retrieval.

[0088] The channel determination results of the candidate information channels can be in the following two situations.

[0089] In the first situation, the channel determination result of the candidate information channel can be information indicating use or non-use. For example, if the channel determination result output by the channel determination model for the candidate information channel is 1, it means using this candidate information channel, and if the channel determination result is 0, it means not using this candidate information channel.

[0090] In this case, the data platform can select, from the channel determination results of each candidate information channel, the information channel whose channel determination result is indicated to be used as the target information channel.

[0091] In the second case, the channel determination result of the candidate information channel is represented by the probability of using the candidate information channel.

[0092] In this case, the candidate information channels can be sorted according to the descending order of each channel determination result, and one or more information channels with the highest ranking can be selected as the target information channels.

[0093] The above channel determination model is a trained model with channel determination ability. By using the channel determination model, the target information channel can be accurately determined, and thus data retrieval can be performed based on the target information channel, which can improve the accuracy of data retrieval.

[0094] On the basis of the foregoing embodiments, after the data platform obtains the channel determination results of each candidate information channel, it can also determine the target information channel in the manner mentioned in the following embodiments.

[0095] In an embodiment of the present disclosure, the data platform determines the initial information channels among the candidate information channels according to the channel determination results; and determines the target information channels among the initial information channels whose retrieval priorities are higher than the preset priority threshold according to the preset retrieval priorities of the candidate information channels.

[0096] Among them, the above retrieval priority is the priority of using each candidate information channel for data retrieval during the data retrieval process.

[0097] The above retrieval priority can be set manually by the user or determined according to the recommendation coefficient of each candidate information channel. For example, the more professional, the higher the timeliness, and the better the user feedback of the data provided by the candidate information channel, the higher the recommendation coefficient of the candidate information channel and the higher the retrieval priority.

[0098] The above priority threshold is a preset threshold.

[0099] For the method of determining the initial information channels, reference can be made to the foregoing embodiments, which will not be elaborated here.

[0100] After determining the initial information channels, the target information channels among the initial information channels whose retrieval priorities are higher than the preset priority threshold can be determined according to the preset retrieval priorities of the candidate information channels.

[0101] For example, the data platform can determine that the information channels among the initial information channels whose numerical values representing the priority are higher than the preset numerical value are the target information channels.

[0102] For another example, the data platform may determine that the information channels with the priority ranking numbers in the top preset number of positions in each initial information channel are the target information channels.

[0103] As can be seen from the above, when retrieving data using the solution provided by the embodiments of the present disclosure, the data platform not only performs a first screening from each candidate information channel according to the channel determination result, but also performs a second screening according to the retrieval priorities of each candidate information channel. In this way, by using the two screenings, the target information channel can be accurately screened out, and thus data retrieval can be performed based on the target information channel, which can improve the accuracy of data retrieval.

[0104] In the existing data retrieval solutions, when the data platform retrieves data in multiple information channels and obtains multiple channel retrieval results, it directly feeds back the obtained multiple channel retrieval results to the user. Since the formats of the results output by different information channels may be different, after obtaining multiple channel retrieval results, the user still needs to process the retrieval results of each channel by himself to obtain valuable data.

[0105] To solve the above technical problems, in an embodiment of the present disclosure, each candidate information channel is set with a data format of the output data, and the data formats of the output data set by different candidate information channels may be different or the same.

[0106] In this case, according to the intent parameter, the data platform can not only determine the target information channel, but also determine the first data format of the data retrieval result to be fed back to the user.

[0107] The implementation manner of determining the first data format will be described below.

[0108] The data platform may input the intent parameter into the trained format determination model, and obtain the data format determined by the format determination model according to the intent parameter as the first data format.

[0109] In addition, since each candidate information channel is set with a data format of the output data, when each candidate information channel outputs the retrieval result, it can output the retrieval result in the set data format. Therefore, when the data platform obtains the channel retrieval result fed back by the target information channel, by inputting the channel input information into the target information channel for data retrieval, it can obtain the channel retrieval result in the second data format fed back by the target information channel.

[0110] Wherein, the above second data format is: the data format of the output data set by the target information channel.

[0111] After the data platform obtains the channel retrieval results in the second data format and determines the first data format of the data retrieval results, during the process of data fusion for each channel retrieval result, it can perform format conversion on each channel retrieval result according to the conversion relationship between the first data format and the second data format to obtain the channel retrieval results in the first data format, and then perform data fusion on the channel retrieval results after format conversion to obtain the data fusion result.

[0112] As can be seen from the above, when retrieving data using the solution provided by the embodiments of the present disclosure, since the intent parameter represents the user's retrieval intent, therefore, according to the intent parameter, the first data format of the data retrieval results that conforms to the user's intent can be accurately determined. During the data fusion process, according to the conversion relationship between the first data format and the second data format, the channel retrieval results in the second data format are converted into the channel retrieval results in the first data format, and then fused, so that the data fusion result in the first data format can be obtained. Using the data fusion result in the first data format as the data retrieval result and feeding it back to the user can feed back the data retrieval result in a data format that conforms to the user's intent to the user.

[0113] Moreover, since the data platform has performed data fusion on each channel retrieval result, there is no need for the user to process each channel retrieval result anymore. Therefore, applying the data retrieval solution provided by the embodiments of the present disclosure can reduce the workload of the user, reduce unnecessary post-processing links for the user, and improve the user experience.

[0114] The data platform can provide data retrieval services for users in different fields (or different business scenarios), such as education, medical, sports and other fields. In the prior art, for each field, or each business scenario, the data platform needs to pre-understand the data characteristics of the business scenario and set the data retrieval method for retrieving data in each information channel according to the data characteristics of the business scenario. It can be seen that the deployment difficulty of the existing data retrieval solution is relatively large. Moreover, in the case of a new scenario, the data platform needs to understand the data characteristics of the new scenario again and adapt the data retrieval methods corresponding to each existing information channel according to the data characteristics of the new scenario. Therefore, the existing data retrieval solution has poor scalability and is difficult to flexibly cope with the dynamic changes of the scenario, and the repeated understanding of the data characteristics of the new scenario results in a waste of resources.

[0115] To solve the above technical problems, in an embodiment of the present disclosure, the data platform also sets the data format of the input data for each candidate information channel, and the data formats of the input data set for different candidate information channels can be different or the same.

[0116] In this case, for each target information channel, when the data platform obtains the channel input information corresponding to the target information channel, it can extract the keywords carried in the data retrieval request, and obtain the channel input data for data retrieval input to the target information channel according to the extracted keywords and the third data format.

[0117] Among them, the third data format is: the data format of the input data set by the target information channel.

[0118] Specifically, the data platform can use keyword extraction technology to extract the keywords carried in the data retrieval request, and set the format of the keywords to the third data format to obtain the keywords in the third data format, which are used as the channel input data for data retrieval input to the target information channel.

[0119] As can be seen from the above, when retrieving data using the solution provided by the embodiments of the present disclosure, each candidate information channel is set with a fixed data format of the input data. In this way, no matter which business scenario the data retrieval request comes from, the data retrieval request is processed in the same way, which can form a unified data processing framework, that is, extract the keywords carried in the data retrieval request, obtain the channel input data according to the keywords and the third data format, and then input the channel input data into the channel for data retrieval. It can be seen that the solution provided by the embodiments of the present disclosure forms a unified processing framework, with low deployment difficulty, and can be adapted to multiple scenarios. It can not only meet the universality of data retrieval, but also be applied to data retrieval in different business scenarios, so as to realize differentiated information capture and personalized output functions for different scenario requirements.

[0120] Moreover, the processing flow of the data retrieval request in different business scenarios of this solution remains unchanged, and each step in the process can adopt a modular design, that is, each step is processed by a module. Therefore, this solution can be quickly deployed and iteratively upgraded in a variety of different business scenarios.

[0121] For the case of new scenarios emerging, since the data platform can process the data retrieval requests in any business scenario in the same way, even if new scenarios emerge, the data platform does not need to understand the data characteristics of the new scenarios again, so that the access of the new scenarios can be completed at the lowest cost. Therefore, applying the data retrieval solution provided by the embodiments of the present disclosure can reduce redundant development costs, save resources, improve data retrieval efficiency, shorten the iterative update cycle of data retrieval, improve the scalability of data retrieval, and can also flexibly respond to the dynamic changes of business scenarios, ensuring the stability and continuity of data retrieval, and providing strong technical support and stable guarantee for enterprises in a dynamically changing business environment.

[0122] In one embodiment of the present disclosure, each candidate information channel of the data platform is set with the data formats of the input data and the output data. When performing data retrieval in the candidate information channel, it is necessary to generate the channel input data according to the set data format of the input data, and the retrieval result output by the candidate information channel is the result in the data format of the set output data. This can abstract the candidate information channel into a standardized data interface and achieve accurate data acquisition.

[0123] The data platform can also process data in the manner mentioned in the following embodiments.

[0124] In one embodiment of the present disclosure, the data platform can determine the target classification type of the data that the user is interested in according to the intent parameter. In this way, when obtaining the channel retrieval result feedback by the target information channel, the channel input information can be input into the target information channel for data retrieval to obtain the initial retrieval result feedback by the target information channel, and the data belonging to the target classification type in the initial retrieval result can be screened out as the channel retrieval result of the target information channel.

[0125] Among them, the classification types of the data can include sports, education, medical treatment, etc.

[0126] Specifically, the data platform can input the intent parameter into the trained type determination model to obtain the data classification type determined by the type determination model according to the intent parameter as the target classification type of the data that the user is interested in.

[0127] On the one hand, the data platform determines the target classification type, and on the other hand, it can input the channel input information into the target information channel for data retrieval to obtain the initial retrieval result feedback by the target information channel. After determining the target classification type and obtaining the initial retrieval result, the data platform can screen out the data belonging to the target classification type in the initial retrieval result as the channel retrieval result of the target information channel.

[0128] For example, if the target classification type of the data that the user is interested in is sports, the data platform can screen out the sports data in the initial retrieval result as the channel retrieval result of the target information channel.

[0129] As can be seen from the above, when retrieving data using the solution provided by the embodiments of the present disclosure, the intent parameter characterizes the user's intent. Based on the intent parameter, the data platform can accurately determine the target classification type of the data that the user is interested in. In this way, when the data platform inputs the channel input information into the target information channel for data retrieval, obtains the initial retrieval result feedback by the target information channel, and filters out the data belonging to the target classification type from the initial retrieval result, it can obtain the data that meets the user's intent, or the data that the user is interested in, as the channel retrieval result. Furthermore, by fusing the data of each channel retrieval result and feeding back the data fusion result to the user, it can provide the user with the data that the user is interested in, thereby improving the accuracy of data retrieval and enhancing the user experience.

[0130] When the data platform obtains the intent parameter and the channel input information, it can adopt the methods mentioned in the following embodiments.

[0131] In an embodiment of the present disclosure, when the data platform obtains the intent parameter, it can obtain the intent parameter characterizing the user's retrieval intent according to the data retrieval request and the intent acquisition strategy; when obtaining the channel input information, it can obtain the channel input information input into the target information channel for data retrieval according to the data retrieval request and the information acquisition strategy.

[0132] Among them, the intent acquisition strategy and / or the information acquisition strategy are dynamically adjusted according to at least one of the following information:

[0133] The user's historical retrieval information, the data change information characterizing the data change situation in the candidate information channels.

[0134] First, the historical retrieval information and the data change information are introduced.

[0135] 1. Historical retrieval information

[0136] The user's historical retrieval information includes various information obtained by the data platform during the historical data retrieval process, such as the intent information, channel input information, channel retrieval result, and data fusion result during the historical data retrieval process, and so on.

[0137] 2. Data change information

[0138] The data change information characterizes the data change situation in the candidate information channels, and this data change situation can refer to the change situation of the data within a preset time period.

[0139] There are the following two types of this data change situation.

[0140] The first type: The data change situation is the change situation of the proportion of various types of data in the candidate information channels.

[0141] For example, the candidate information channel may contain three types of data, namely type 1, type 2, and type 3. The proportion distribution of these three types of data at a certain moment is 5:2:3. After a time period, the proportion distribution of these three types of data is 4:5:1. It can be seen that in the candidate information channel, the data of type 2 increases, while the data of type 1 and type 3 decreases, indicating that the data of type 2 is more popular, and the data of type 1 and type 3 is not popular. At this time, the data change information is the information indicating that the data of type 2 increases and the data of type 1 and type 3 decreases.

[0142] Second: The data change situation is the change situation of the proportion of data of each classification type retrieved during the historical data retrieval process of the candidate information channel.

[0143] For example, for the data of the same classification type, among all the data retrieved within a historical period, the proportion of the data of this classification type is 0.5, and the proportion in the next historical period is 0.3. Then the proportion of the data of this classification type has decreased by 0.2, indicating that the user's interest in the data of this classification type has decreased. At this time, the data change information is the information indicating the decrease in the proportion of the data of this classification type.

[0144] The process of adjusting the intention acquisition strategy will be described below.

[0145] Historical retrieval information can be understood as which data the user has retrieved during the historical retrieval process, which can illustrate which data the user is interested in. In this way, based on the historical retrieval information, the preference or habit of the user when retrieving data can be determined. Therefore, by adjusting the intention acquisition strategy according to the historical retrieval information, when the data platform obtains the intention parameters according to the intention acquisition strategy next time, it can obtain the intention parameters that conform to the user's preference or habit.

[0146] For example, based on the historical retrieval information, the data platform can determine that the user is interested in picture data. Then the intention acquisition strategy can be adjusted so that when the data platform obtains the intention parameters according to the intention acquisition strategy next time, it can obtain the intention parameters representing the user's intention to obtain picture data.

[0147] The data change information can also illustrate which data the user is interested in. For example, if the data change information indicates that the proportion of picture data increases, it means that the user is more interested in picture data. In this way, by adjusting the intention acquisition strategy according to the data change details, when the data platform obtains the intention parameters according to the intention acquisition strategy next time, it can obtain the intention parameters representing the user's intention to obtain picture data.

[0148] The process of adjusting the information acquisition strategy will be described below.

[0149] The above historical retrieval information and data change information can illustrate which data the user is interested in. Adjusting the information acquisition strategy according to at least one of these two types of information enables the data platform to obtain, as the channel input information when obtaining channel input information according to the information acquisition strategy next time, the type of data that the user is interested in and is included in the data retrieval request.

[0150] As can be seen from the above, when retrieving data using the solution provided by the embodiments of the present disclosure, the intent parameter is obtained according to the intent acquisition strategy, and the channel data information is obtained according to the information acquisition strategy, and at least one of these two strategies is dynamically adjusted according to the historical retrieval information and / or data change information. In this way, by dynamically adjusting the strategy, the data retrieval process is optimized in real time, the adaptability and scalability of the present solution are improved, and the long-term stable operation of the present solution in the data processing of multiple scenarios, multiple channels, and multiple data types is ensured. And dynamically adjusting these two strategies can improve the timeliness of these two strategies. Therefore, retrieving data according to these two strategies can improve the accuracy of data retrieval and ensure the precise matching of the data feedback by the data platform with the user's intent and the efficient transmission of data.

[0151] Corresponding to the foregoing data retrieval method, the embodiments of the present disclosure further provide a data retrieval device.

[0152] In an embodiment of the present disclosure, referring to Figure 3 , a structural schematic diagram of a data retrieval device based on artificial intelligence is provided. In this embodiment, the above device includes:

[0153] An intent acquisition module 301, configured to, in response to receiving a data retrieval request, obtain an intent parameter characterizing the retrieval intent of the user according to the data retrieval request;

[0154] A channel determination module 302, configured to determine a target information channel that matches the intent parameter among the candidate information channels;

[0155] An information acquisition module 303, configured to obtain, according to the data retrieval request, channel input information input to the target information channel for data retrieval;

[0156] A result acquisition module 304, configured to obtain a channel retrieval result of inputting the channel input information to the target information channel for data retrieval;

[0157] A result feedback module 305, configured to perform data fusion on the obtained channel retrieval results and feedback the obtained fusion result as the data retrieval result.

[0158] As can be seen from the above, when retrieving data using the solution provided by the embodiments of the present disclosure, the data platform can accurately determine a target information channel that matches the intention parameter representing the user's retrieval intention according to the intention parameter. In this way, without the user specifying the information channel, the information channel that meets the user's intention can be determined, and thus data retrieval can be performed in the target information channel to obtain a channel retrieval result that meets the user's intention. By fusing the retrieval results of each channel and using the obtained fusion result as the data retrieval result for feedback, data that meets the user's intention can be fed back to the user. Therefore, by applying the data retrieval solution provided by the embodiments of the present disclosure, the accuracy of data retrieval can be improved, and the user experience can be enhanced.

[0159] In one embodiment of the present disclosure, the intention acquisition module 301 includes:

[0160] A parameter detection sub-module, configured to detect whether an intention parameter representing the user's retrieval intention is carried in the data retrieval request. If not, the intention recognition sub-module is triggered;

[0161] The intention recognition sub-module is configured to perform intention recognition based on the data retrieval request to obtain the intention parameter.

[0162] In this way, the data platform can quickly obtain the intention parameter when the data retrieval request includes the intention parameter, and thus perform data retrieval based on the intention parameter, which can improve the efficiency of data retrieval.

[0163] In one embodiment of the present disclosure, the intention recognition sub-module is specifically configured to:

[0164] Input the parameters carried in the data retrieval request into a trained intention recognition model, and obtain the intention recognition result output by the intention recognition model as the intention parameter.

[0165] The above-mentioned intention recognition model is a trained model with intention recognition ability. Using the intention recognition model for intention recognition can accurately obtain the intention parameter, and thus perform data retrieval based on the intention parameter, which can improve the accuracy of data retrieval.

[0166] In one embodiment of the present disclosure, the channel determination module 302 includes:

[0167] A channel determination sub-module, configured to input the intention parameter into a trained channel determination model, and obtain the channel determination results of each candidate information channel output by the channel determination model, where the channel determination result is: the result determined by the channel determination model based on the intention parameter, representing whether to use the candidate information channel for data retrieval;

[0168] A channel determination sub-module, configured to determine a target information channel among each candidate information channel according to the channel determination result.

[0169] The above-mentioned channel determination model is a trained model with channel determination ability. By using the channel determination model, the target information channel can be accurately determined, and thus based on the target information channel for data retrieval, the accuracy of data retrieval can be improved.

[0170] In an embodiment of the present disclosure, the channel determination sub-module is specifically configured to:

[0171] Determine an initial information channel among each candidate information channel according to the channel determination result;

[0172] Determine a target information channel among each initial information channel whose retrieval priority is higher than a preset priority threshold according to the preset retrieval priorities of each candidate information channel.

[0173] As can be seen from the above, when retrieving data using the solution provided by the embodiment of the present disclosure, the data platform not only performs a first screening from each candidate information channel according to the channel determination result, but also performs a second screening according to the retrieval priorities of each candidate information channel. In this way, by using the two screenings, the target information channel can be accurately screened out, and thus based on the target information channel for data retrieval, the accuracy of data retrieval can be improved.

[0174] In an embodiment of the present disclosure, each candidate information channel is set with a data format of the output data;

[0175] The apparatus further includes:

[0176] A format determination module, configured to determine a first data format of the data retrieval result according to the intention parameter;

[0177] The result obtaining module 304 is specifically configured to:

[0178] Input the channel input information into the target information channel for data retrieval, and obtain a channel retrieval result in a second data format fed back by the target information channel, where the second data format is: the data format of the output data set by the target information channel;

[0179] The result feedback module 305 is specifically configured to:

[0180] Perform format conversion on each channel retrieval result according to the conversion relationship between the first data format and the second data format to obtain each channel retrieval result in the first data format;

[0181] Perform data fusion on each channel retrieval result after format conversion to obtain the data fusion result.

[0182] As can be seen from the above, when retrieving data using the solution provided in the embodiments of the present disclosure, since the intent parameter characterizes the user's retrieval intent, therefore, according to the intent parameter, the first data format of the data retrieval result that conforms to the user's intent can be accurately determined. During the data fusion process, according to the conversion relationship between the first data format and the second data format, the channel retrieval result in the second data format is converted into the channel retrieval result in the first data format, and then fused. In this way, the data fusion result in the first data format can be obtained, and the data fusion result in the first data format is used as the data retrieval result and fed back to the user, which can feed back the data retrieval result in the data format that conforms to the user's intent to the user.

[0183] Moreover, since the data platform has performed data fusion on the channel retrieval results of each channel, there is no need for the user to process the channel retrieval results of each channel anymore. Therefore, by applying the data retrieval solution provided in the embodiments of the present disclosure, the workload of the user can be reduced, unnecessary post-processing links of the user are reduced, and the user experience is improved.

[0184] In one embodiment of the present disclosure, the apparatus further includes:

[0185] a type determination module, configured to determine a target classification type of data that the user is interested in according to the intent parameter;

[0186] The result obtaining module 304 is specifically configured to:

[0187] input the channel input information into the target information channel for data retrieval to obtain an initial retrieval result fed back by the target information channel;

[0188] screen out the data belonging to the target classification type from the initial retrieval result as the channel retrieval result of the target information channel.

[0189] As can be seen from the above, when retrieving data using the solution provided in the embodiments of the present disclosure, the intent parameter characterizes the user's intent. The data platform can accurately determine the target classification type of the data that the user is interested in according to the intent parameter. In this way, when the data platform inputs the channel input information into the target information channel for data retrieval, obtains the initial retrieval result fed back by the target information channel, and screens out the data belonging to the target classification type from the initial retrieval result, the data that conforms to the user's intent, or the data that the user is interested in, can be obtained as the channel retrieval result. Furthermore, data fusion is performed on the channel retrieval results of each channel, and the data fusion result is fed back to the user, which can provide the data that the user is interested in for the user, thereby improving the accuracy of data retrieval and enhancing the user experience.

[0190] In one embodiment of the present disclosure, each candidate information channel is set with the data format of the input data;

[0191] The information acquisition module 303 is specifically configured to:

[0192] Extract the keywords carried in the data retrieval request, and obtain the channel input data for data retrieval input to the target information channel according to the extracted keywords and the third data format, where the third data format is the data format of the input data set by the target information channel.

[0193] As can be seen from the above, when retrieving data using the solution provided by the embodiments of the present disclosure, each candidate information channel is set with a fixed data format of the input data. In this way, no matter which business scenario the data retrieval request comes from, the same method is used to process the data retrieval request, which can form a unified data processing framework, that is, extract the keywords carried in the data retrieval request, obtain the channel input data according to the keywords and the third data format, and then input the channel input data into the channel for data retrieval. It can be seen that the solution provided by the embodiments of the present disclosure forms a unified processing framework, with low deployment difficulty, enabling adaptation to multiple scenarios, being able to meet the universality of data retrieval, and being applicable to data retrieval in different business scenarios, thereby realizing the differential information capture and personalized output functions for different scenario requirements.

[0194] Moreover, the processing flow of the data retrieval request in different business scenarios of this solution remains unchanged, and each step in the process can adopt a modular design, that is, each step is processed by a module. Therefore, this solution can be quickly deployed and iteratively upgraded in multiple different business scenarios.

[0195] For the situation of newly added scenarios, since the data platform can process the data retrieval requests in any business scenario in the same way, even if a new scenario appears, the data platform does not need to re-understand the data characteristics of the new scenario, enabling the access of the new scenario to be completed at the lowest cost. Therefore, applying the data retrieval solution provided by the embodiments of the present disclosure can reduce redundant development costs, save resources, improve data retrieval efficiency, shorten the iterative update cycle of data retrieval, improve the scalability of data retrieval, and can also flexibly respond to the dynamic changes of business scenarios, ensuring the stability and continuity of data retrieval, providing strong technical support and stable guarantee for enterprises in a dynamically changing business environment.

[0196] In an embodiment of the present disclosure, the result feedback module 305 is specifically configured to:

[0197] Input the retrieved results of each channel obtained into the trained data fusion model to obtain the data fusion result of the data fusion output of the retrieved results of each channel by the data fusion model.

[0198] After the data platform obtains the retrieval results of each channel, it inputs the retrieval results of each channel into the data fusion model. The data fusion model can accurately fuse the data of the retrieval results of each channel and output a relatively accurate data fusion result. Then, subsequent operations are performed based on the data fusion result, which can improve the accuracy of data retrieval.

[0199] In one embodiment of the present disclosure, the intent acquisition module 301 is specifically configured to:

[0200] According to the data retrieval request, obtain intent parameters representing the user's retrieval intent according to the intent acquisition strategy;

[0201] The information acquisition module 303 is specifically configured to:

[0202] According to the data retrieval request, obtain the channel input information input to the target information channel for data retrieval according to the information acquisition strategy;

[0203] Wherein, the intent acquisition strategy and / or the information acquisition strategy are dynamically adjusted according to at least one of the following information:

[0204] The user's historical retrieval information, the data change information representing the data change situation in the candidate information channel.

[0205] As can be seen from the above, when retrieving data using the solution provided in the embodiment of the present disclosure, the intent parameters are obtained according to the intent acquisition strategy, and the channel data information is obtained according to the information acquisition strategy. At least one of these two strategies is dynamically adjusted according to the historical retrieval information and / or the data change information. In this way, by dynamically adjusting the strategy, the data retrieval process is optimized in real time, the adaptability and scalability of the present solution are improved, and the long-term stable operation of the present solution in the data processing of multiple scenarios, multiple channels, and multiple data types is ensured. And dynamically adjusting these two strategies can improve the timeliness of these two strategies. Therefore, retrieving data based on these two strategies can improve the accuracy of data retrieval and ensure the precise matching of the data feedback by the data platform with the user's intent and the efficient transmission of data.

[0206] In one embodiment of the present disclosure, the device further includes:

[0207] An input parameter verification module, configured to perform input parameter verification on the data retrieval request before obtaining the intent parameters representing the user's retrieval intent according to the data retrieval request. If the verification passes, the intent acquisition module 301 is triggered.

[0208] As can be seen from the above, when retrieving data using the solution provided by the embodiments of the present disclosure, before obtaining the intent parameter characterizing the user's retrieval intent according to the data retrieval request, parameter verification of the input parameters of the data retrieval request can be performed, and subsequent operations are only carried out when the verification passes. This can reduce the workload of the data platform and improve the accuracy of data retrieval.

[0209] In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision, and disclosure of the user's personal information involved all comply with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0210] It should be noted that the head model in this embodiment is not a head model for a specific user and does not reflect the personal information of a specific user.

[0211] It should be noted that the two-dimensional face images in this embodiment are from a public dataset.

[0212] In an embodiment of the present disclosure, an intelligent agent of artificial intelligence is also provided, and the intelligent agent is configured to execute the data retrieval method mentioned in any of the foregoing method embodiments.

[0213] According to the embodiments of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0214] Figure 4 The schematic block diagram of an example electronic device 400 that can be used to implement the embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are only examples and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0215] As Figure 4 shown, the device 400 includes a computing unit 401, which can execute various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 402 or the computer program loaded from the storage unit 408 into the random access memory (RAM) 403. In the RAM 403, various programs and data required for the operation of the device 400 can also be stored. The computing unit 401, the ROM 402, and the RAM 403 are connected to each other through a bus 404. The input / output (I / O) interface 405 is also connected to the bus 404.

[0216] Multiple components in device 400 are connected to I / O interface 405, including: input unit 406, such as a keyboard, mouse, etc.; output unit 407, such as various types of displays, speakers, etc.; storage unit 408, such as a disk, optical disc, etc.; and communication unit 409, such as a network card, modem, wireless communication transceiver, etc. Communication unit 409 allows device 400 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0217] Computing unit 401 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of computing unit 401 include but are not limited to a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Computing unit 401 executes the various methods and processes described above, such as the data retrieval method. For example, in some embodiments, the data retrieval method can be implemented as a computer software program that is tangibly embodied in a machine-readable medium, such as storage unit 408. In some embodiments, part or all of the computer program can be loaded and / or installed onto device 400 via ROM 402 and / or communication unit 409. When the computer program is loaded into RAM 403 and executed by computing unit 401, one or more steps of the data retrieval method described above can be executed. Alternatively, in other embodiments, computing unit 401 can be configured to execute the data retrieval method by any other suitable means (e.g., by means of firmware).

[0218] The various embodiments of the systems and techniques described above in this document can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor, and can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0219] The program code for implementing the methods of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the program codes are executed by the processor or controller, the functions / operations specified in the flowchart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0220] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0221] In order to provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and the input received from the user can be in any form (including acoustic input, voice input, or tactile input).

[0222] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected to each other by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), and the Internet.

[0223] A computer system can include a client and a server. The client and the server are generally far from each other and typically interact through a communication network. The client-server relationship is created by computer programs that run on the respective computers and have a client-server relationship with each other. The server can be a cloud server, a server of a distributed system, or a server that incorporates a blockchain.

[0224] It should be understood that the various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recited in this disclosure can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not limited herein.

[0225] The above specific implementation manners do not constitute a limitation on the protection scope of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the protection scope of this disclosure.

Claims

1. An artificial intelligence-based data retrieval method, comprising: In response to receiving a data retrieval request, obtaining an intention parameter characterizing the user's retrieval intention according to the data retrieval request; Determining a target information channel among candidate information channels that matches the intention parameter; Obtaining channel input information input to the target information channel for data retrieval according to the data retrieval request; Obtaining a channel retrieval result of inputting the channel input information into the target information channel for data retrieval; Performing data fusion on the obtained channel retrieval results, and feeding back the obtained fusion result as the data retrieval result.

2. The method according to claim 1, wherein The obtaining an intention parameter characterizing the user's retrieval intention according to the data retrieval request includes: Detecting whether the data retrieval request carries an intention parameter characterizing the user's retrieval intention; If not, performing intention recognition based on the data retrieval request to obtain the intention parameter.

3. The method according to claim 2, wherein The performing intention recognition based on the data retrieval request to obtain the intention parameter includes: Inputting the parameters carried by the data retrieval request into a trained intention recognition model to obtain an intention recognition result output by the intention recognition model as the intention parameter.

4. The method according to any one of claims 1 to 3, wherein The determining a target information channel among candidate information channels that matches the intention parameter includes: Inputting the intention parameter into a trained channel determination model to obtain a channel determination result of each candidate information channel output by the channel determination model, where the channel determination result is a result of the channel determination model determining whether to use the candidate information channel for data retrieval based on the intention parameter; Determining the target information channel among each candidate information channel according to the channel determination result.

5. The method according to claim 4, wherein The determining the target information channel among each candidate information channel according to the channel determination result includes: Determining an initial information channel among each candidate information channel according to the channel determination result; Determining a target information channel among the initial information channels with a retrieval priority higher than a preset priority threshold according to the preset retrieval priorities of each candidate information channel.

6. The method according to any one of claims 1-3, wherein, Each candidate information channel is set with a data format of output data; The method further includes: Determining a first data format of the data retrieval result according to the intention parameter; The obtaining a channel retrieval result of inputting the channel input information into the target information channel for data retrieval includes: Inputting the channel input information into the target information channel for data retrieval to obtain a channel retrieval result in a second data format fed back by the target information channel, where the second data format is the data format of the output data set by the target information channel; The performing data fusion on the obtained channel retrieval results includes: Performing format conversion on the channel retrieval results according to the conversion relationship between the first data format and the second data format to obtain the channel retrieval results in the first data format; Performing data fusion on the channel retrieval results after format conversion.

7. The method according to any one of claims 1-3, wherein, The method further includes: Determining a target classification type of data of interest to the user according to the intention parameter; Obtaining a channel retrieval result by inputting the channel input information into the target information channel for data retrieval includes: Inputting the channel input information into the target information channel for data retrieval to obtain an initial retrieval result fed back by the target information channel; Screening out the data belonging to the target classification type from the initial retrieval result as the channel retrieval result of the target information channel.

8. The method according to any one of claims 1-3, wherein Each candidate information channel is set with a data format of input data; Obtaining the channel input information input into the target information channel for data retrieval according to the data retrieval request includes: Extracting the keywords carried in the data retrieval request, and obtaining the channel input data input into the target information channel for data retrieval according to the extracted keywords and the third data format, where the third data format is the data format of the input data set by the target information channel.

9. The method according to any one of claims 1-3, wherein Performing data fusion on the obtained channel retrieval results includes: Inputting the obtained channel retrieval results into a trained data fusion model to obtain a fusion result output by the data fusion model for data fusion of the channel retrieval results.

10. The method according to claim 1, wherein, Obtaining an intention parameter characterizing the user's retrieval intention according to the data retrieval request includes: Obtaining an intention parameter characterizing the user's retrieval intention according to the data retrieval request according to an intention acquisition strategy; Obtaining the channel input information input into the target information channel for data retrieval according to the data retrieval request includes: Obtaining the channel input information input into the target information channel for data retrieval according to the data retrieval request according to an information acquisition strategy; Wherein, the intention acquisition strategy and / or the information acquisition strategy are dynamically adjusted according to at least one of the following information: The user's historical retrieval information, data change information characterizing the data change situation in the candidate information channels.

11. According to the method according to any one of claims 1-3, wherein, Before obtaining an intention parameter characterizing the user's retrieval intention according to the data retrieval request, the method further includes: Performing input parameter verification on the data retrieval request; If the verification passes, execute the step of obtaining an intention parameter characterizing the user's retrieval intention according to the data retrieval request.

12. An artificial intelligence-based data retrieval device includes: An intention acquisition module, configured to, in response to receiving a data retrieval request, obtain an intention parameter characterizing the user's retrieval intention according to the data retrieval request; A channel determination module, configured to determine a target information channel in each candidate information channel that matches the intention parameter; An information acquisition module, configured to obtain the channel input information input into the target information channel for data retrieval according to the data retrieval request; A result acquisition module, configured to obtain a channel retrieval result by inputting the channel input information into the target information channel for data retrieval; A result feedback module, configured to perform data fusion on the obtained channel retrieval results and feed back the obtained fusion result as a data retrieval result.

13. An artificial intelligence agent configured to execute the method according to any one of claims 1-11.

14. An electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method according to any one of claims 1-11.

15. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to execute the method according to any one of claims 1-11.

16. A computer program product, comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1-11.

Images