Retrieval processing method and device based on large model and barrier-free service

Abstract

The invention provides a retrieval processing method and device based on a large model and barrier-free service, and relates to the field of artificial intelligence such as large models, mobile internet, computer vision and cloud computing. The method comprises the steps that first description information is obtained, M barrier-free services are called respectively, M is a positive integer larger than 1, the first description information is description information capable of determining a corresponding target commodity, and the target commodity is a commodity to be retrieved by a user; and for any barrier-free service, performing the following processing: obtaining a retrieval result returned by the barrier-free service, screening out a target retrieval result from the retrieval result by utilizing a large model, and returning the target retrieval result to a user, the barrier-free service is used for performing retrieval in the corresponding e-commerce platform according to the first description information and determining the retrieval result according to a low price priority principle.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to a retrieval processing method and device based on big models and barrier-free services in the fields of big models, mobile Internet, computer vision, and cloud computing. Background Art

[0002] With the continuous development of mobile Internet, major Internet companies have launched their own e-commerce platforms, namely e-commerce applications (apps), and provided different discounts for certain specific products on their own e-commerce platforms. Summary of the invention

[0003] The present disclosure provides a retrieval processing method and device based on a large model and barrier-free service.

[0004] A retrieval processing method based on a large model and barrier-free service, comprising:

[0005] Obtaining first description information, and calling M accessibility services respectively, where M is a positive integer greater than 1, and the first description information is description information capable of determining a corresponding target product, and the target product is a product to be searched by the user;

[0006] For any accessibility service, the following processing is performed respectively: the search results returned by the accessibility service are obtained, and the target search results are filtered out from the search results using a large model, and returned to the user. The accessibility service is used to search the corresponding e-commerce platform according to the first description information and determine the search results according to the principle of low price first.

[0007] A retrieval processing device based on a large model and barrier-free service comprises: a first processing module and a second processing module;

[0008] The first processing module is used to obtain first description information and call M accessibility services respectively, where M is a positive integer greater than 1, and the first description information is description information capable of determining a corresponding target product, and the target product is a product to be searched by the user;

[0009] The second processing module is used to perform the following processing for any accessibility service: obtain the search results returned by the accessibility service, and use the big model to filter out target search results from the search results, and return them to the user. The accessibility service is used to search in the corresponding e-commerce platform according to the first description information and determine the search results according to the principle of low price first.

[0010] An electronic device, comprising:

[0011] at least one processor; and

[0012] a memory communicatively connected to the at least one processor; wherein,

[0013] The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the at least one processor to enable the at least one processor to perform the method described above.

[0014] A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute the method as described above.

[0015] A computer program product comprises a computer program / instruction, wherein the computer program / instruction implements the method described above when executed by a processor.

[0016] 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 intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0018] Figure 1 A flowchart of an embodiment of a retrieval processing method based on a large model and barrier-free service according to the present disclosure;

[0019] Figure 2 A schematic diagram of the interaction between the user, access service, proxy service and barrier-free service described in the present disclosure;

[0020] Figure 3 It is a schematic diagram of the composition structure of an embodiment 300 of a retrieval processing device based on a large model and barrier-free service according to the present disclosure;

[0021] Figure 4 A schematic block diagram of an electronic device 400 that can be used to implement an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0022] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0023] In addition, it should be understood that the term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0024] Figure 1 Flow chart of an embodiment of the retrieval processing method based on a large model and barrier-free service described in the present disclosure. Figure 1 As shown, the following specific implementation methods are included.

[0025] In step 101, first description information is obtained, and M accessibility services are called respectively, where M is a positive integer greater than 1. The first description information is description information that can determine the corresponding target product, and the target product is the product to be searched by the user.

[0026] In step 102, for any accessibility service, the following processing is performed respectively: the search results returned by the accessibility service are obtained, and the target search results are filtered out from the search results using the large model and returned to the user. The accessibility service is used to search the corresponding e-commerce platform according to the first description information and determine the search results according to the principle of low price first.

[0027] As mentioned earlier, major Internet companies have launched their own e-commerce platforms and can provide different discounts for certain specific products on their own e-commerce platforms. This makes it impossible for users to find the products they want through a unified search portal.

[0028] By adopting the scheme described in the above method embodiment, after obtaining the first description information, M accessibility services can be called respectively, so that the M accessibility services can be used to retrieve the search results corresponding to the first description information from the corresponding e-commerce platforms respectively, and the search results from each e-commerce platform can be screened using a large model and returned to the user, that is, the searches of each e-commerce platform are unified into one entrance, so that users can directly obtain the same product information on different e-commerce platforms, which is convenient for users to compare prices, etc., thereby simplifying user operations and improving users' information acquisition efficiency. Moreover, by screening the search results, users' information acquisition efficiency is further improved and the quality of information obtained by users is improved.

[0029] In practical applications, Figure 1The execution subject of the illustrated embodiment may be an agent service, and the agent service may have its own unique identifier (ID). The agent service may correspond to multiple accessibility services, and the specific number may be determined according to actual needs. The accessibility service may be understood as an app that is pre-developed and runs on the simulator service, which can actually execute the search command and obtain the search results. There may be a one-to-one relationship between the simulator service and the accessibility service. When the accessibility service is started, it will query the correspondence between the agent service and the simulator service, and establish a connection with the agent service through the websocket protocol. Each accessibility service may correspond to an e-commerce platform.

[0030] In some embodiments of the present disclosure, a method for obtaining the first description information may include: obtaining a first call request from an access service, and obtaining the first description information from the first call request, where the first description information is description information input by a user through an interactive interface.

[0031] The user can input the first description information in the input box of the interactive interface, such as "People's Education Edition Grade 3 Mathematics Textbook Volume 1", and the corresponding target product can be clearly determined based on the description information. After the user inputs the first description information, a search request can be triggered, and then an access service (ServerLess) can be triggered through a Hypertext Transfer Protocol (HTTP) trigger. The access service is used to call other services according to the user's request. Accordingly, the access service can call the proxy service, such as sending a first call request to the proxy service, which can carry the first description information, so that the proxy service can obtain the first description information.

[0032] In some embodiments of the present disclosure, the method for obtaining the first description information may also include: obtaining a second call request from an access service, and obtaining second description information input by a user through an interactive interface from the second call request, the second description information being description information of a target product for which a corresponding description cannot be determined, calling M accessibility services respectively, and obtaining search results returned by each accessibility service for the second description information respectively, and returning them to the user, determining the product corresponding to the search result selected by the user from each search result as the target product, and determining the description information of the target product generated according to the selected search result as the first description information.

[0033] The user can enter the second description information in the input box of the interactive interface, such as "** brand mobile phone". According to the description information, the corresponding target product cannot be clearly determined, because the ** brand mobile phone includes a variety of different products, and it is impossible to determine which one it refers to. In this case, M barrier-free services can be called separately, such as sending a search request to each barrier-free service. The search request can carry the second description information, and the search results returned by each barrier-free service can be obtained separately (low price priority principle) and returned to the user. Since each barrier-free service usually corresponds to only one e-commerce platform, multiple barrier-free services can be called in parallel. Among them, the number of search results returned by each barrier-free service can be determined according to actual needs. Furthermore, the product corresponding to the search result selected by the user from the returned search results can be determined as the target product, and the description information of the target product generated according to the selected search result can be determined as the first description information. For example, each search result returned may include a product details page image, the product in the product details page image selected by the user (such as a mobile phone of **brand **model) may be identified as the target product, and the first description information may be generated based on the product information in the product details page image selected by the user, and the specific form is not limited.

[0034] The above process can be called a rough search process, that is, when the corresponding target product cannot be clearly determined based on the description information entered by the user, the rough search results can be obtained by calling various accessibility services, and the user can manually select to confirm the target product.

[0035] In addition, the user can directly input the first description information or input the second description information so as to determine the first description information based on the second description information. This is very flexible and convenient, and can meet the different usage requirements of the user.

[0036] After obtaining the first description information, M accessibility services may be called separately, such as sending a search request to each accessibility service. The search request may carry the first description information. Accordingly, each accessibility service searches the corresponding e-commerce platform according to the first description information and determines the search results to be returned based on the principle of low price first. For example, assuming that a certain accessibility service retrieves a total of 100 search results, the 10 search results with the lowest prices may be returned.

[0037] Accordingly, for any barrier-free service, the following processing can be performed respectively: obtain the search results returned by the barrier-free service, and use the big model to filter out the target search results from the search results and return them to the user.

[0038] In some embodiments of the present disclosure, for the search results corresponding to any accessibility service, in response to obtaining the filtering conditions set by the user for the target product, a large model can be used to determine low-quality search results that do not meet the filtering conditions from the search results, and the low-quality search results can be deleted, and then the target search results can be filtered out from the remaining search results.

[0039] The filtering conditions can be set by the user, and the specific setting method is not limited. For example, the filtering conditions can be set while inputting the first description information.

[0040] By deleting low-quality search results, the workload of subsequent processing can be reduced, and processing efficiency can be improved and resource consumption can be reduced.

[0041] In some embodiments of the present disclosure, the filtering conditions may include N sub-conditions, where N is a positive integer. Accordingly, for the search results corresponding to any accessible service, a method of using a large model to determine low-quality search results that do not meet the filtering conditions from the search results may include: for any search result, performing the following processing respectively: using the large model to respectively determine the evaluation results of the search result corresponding to each sub-condition, and in response to determining that the search result does not meet any sub-condition based on the evaluation results, determining the search result as a low-quality search result.

[0042] Specifically, in some embodiments of the present disclosure, any search result may include: merchant details page pictures, product details page pictures and comment details page pictures. In addition, the number of large models may be N, and each sub-condition may correspond to a large model. Accordingly, for any search result, the method of using the large model to determine the evaluation results of the search result corresponding to each sub-condition may include: for any sub-condition, the following processing is performed respectively: from the pictures included in the search result, the picture corresponding to the sub-condition is determined as the target picture, and a prompt word (prompt) is constructed according to the sub-condition, and the target picture and the prompt word are used as the input of the large model corresponding to the sub-condition to obtain the output evaluation result. In theory, all comments need to be included in the comment details page picture, but if the number of comments is large, only the latest generated W comments can be retained, etc. W is a positive integer greater than 1, and the specific value can be determined according to actual needs.

[0043] The specific value of N can also be determined according to actual needs. In addition, the specific contents included in each of the N sub-conditions can also be determined according to actual needs. For example, the value of N can be 4, and the 4 sub-conditions can be: the number of negative reviews is less than the first threshold or the proportion of negative reviews is less than the second threshold (first sub-condition), it comes from a flagship store (second sub-condition), the product is genuine (third sub-condition), and the product in the picture is the target product (fourth sub-condition). The specific values ​​of the first threshold and the second threshold can be determined according to actual needs.

[0044] Taking a certain search result a as an example, for the first sub-condition (assuming that the first sub-condition is that the number of negative reviews is less than the first threshold), the comment details page picture in the pictures included in the search result a can be used as the target picture, and a prompt word can be constructed according to the first sub-condition, such as "identify the number of negative reviews based on the information in the picture", etc., and then the comment details page picture and the constructed prompt word can be used as the input of the large model corresponding to the first sub-condition to obtain the output first evaluation result, which can be the number of negative reviews identified by the large model (possibly 0). For the second sub-condition, the merchant details page picture can be used as the target picture, and the second evaluation result obtained can be whether it is from the flagship store. For the third sub-condition, the comment details page picture can be used as the target picture, and the third evaluation result obtained can be whether it is genuine. For the fourth sub-condition, the product details page picture can be used as the target picture, and the fourth evaluation result obtained can be whether it is the target product, that is, to compare whether the retrieved product is the same as the target product.

[0045] The large model refers to a large language model (LLM), which is a deep learning model trained with a large amount of text data, capable of generating natural language text or understanding the meaning of language text, and can handle a variety of natural language tasks. In the scheme described in the present disclosure, each sub-condition may correspond to a large model, and the specific type of each large model is not limited, as long as the corresponding function can be achieved.

[0046] For any search result, after obtaining the evaluation results of the search result corresponding to each sub-condition, if it is determined according to the evaluation results that the search result does not meet any sub-condition, the search result can be determined as a low-quality search result.

[0047] For example, if it is determined that the number of negative reviews in the first evaluation result is greater than the first threshold, then it can be considered that the search result does not meet the first sub-condition, and the search result can be determined as a low-quality search result. For another example, if it is determined that the product is a fake according to the third evaluation result, then it can be considered that the search result does not meet the third sub-condition, and the search result can be determined as a low-quality search result.

[0048] Through the above processing, the required evaluation results can be obtained efficiently and accurately with the help of large models corresponding to different sub-conditions, and then low-quality search results can be determined and deleted based on the evaluation results, thereby achieving effective filtering of search results returned by accessibility services and improving the quality of remaining search results.

[0049] Furthermore, the remaining search results may be screened out to obtain target search results.

[0050] In some embodiments of the present disclosure, the N sub-conditions may include at least: a first sub-condition in which the number of negative reviews is less than a first threshold or the proportion of negative reviews is less than a second threshold, and the evaluation results corresponding to each sub-condition may include at least: a first evaluation result corresponding to the first sub-condition, the first evaluation result being the number of negative reviews or the proportion of negative reviews. Accordingly, a method for filtering out target search results from the remaining search results may include: sorting the remaining search results in ascending order of the values ​​of the first evaluation result, and in response to determining that the number of remaining search results is less than or equal to P, where P is a positive integer, determining all sorted search results as target search results, and in response to determining that the number of remaining search results is greater than P, determining the search results in the top P positions after sorting as target search results.

[0051] Assume that the value of P is 3, and take the first sub-condition that the number of negative reviews is less than the first threshold as an example, the remaining search results can be sorted in order from small to large according to the number of negative reviews. If it is determined that the number of remaining search results is less than or equal to 3, then the sorted search results can all be determined as target search results. If it is determined that the number of remaining search results is greater than 3, then the search results in the top 3 after sorting can be determined as target search results.

[0052] In addition, in some embodiments of the present disclosure, the method of filtering out target search results from the remaining search results may also include: performing the following processing for any remaining search result: determining a comprehensive score of the search result in combination with the evaluation results of the search result corresponding to each sub-condition, sorting the remaining search results in order from large to small according to the values ​​of the comprehensive score, and in response to determining that the number of remaining search results is less than or equal to P, where P is a positive integer, determining all sorted search results as target search results; and in response to determining that the number of remaining search results is greater than P, determining the search results in the first P positions after sorting as target search results.

[0053] For example, assuming that the value of N is 4 and the value of P is 3, then for each remaining search result, the four evaluation results corresponding to the search result can be converted into corresponding numerical representations in a predetermined manner, and then the products of each numerical representation and the corresponding weight can be obtained respectively, and the sum of each product can be obtained, and the sum is used as the comprehensive score of the search result. After that, the remaining search results can be sorted in order from large to small according to the comprehensive score. If it is determined that the number of remaining search results is less than or equal to 3, then the sorted search results can all be determined as target search results. If it is determined that the number of remaining search results is greater than 3, then the search results in the top 3 after sorting can be determined as target search results.

[0054] No matter which of the above methods is used, the target search results can be filtered out from the remaining search results efficiently and accurately. The specific method to be used can be determined according to actual needs, which is very flexible and convenient.

[0055] For each accessible service / e-commerce platform, after determining the target search results, a result list can be generated. For example, if the number of target search results is 3, then the result list can include these 3 target search results, and the result list can be returned to the user. If the number of accessible services is 4, then the user can obtain 4 result lists, that is, the list of the lowest-priced products that meet the screening conditions on the 4 e-commerce platforms.

[0056] For users, they can simultaneously view information such as the price of the target product on different e-commerce platforms (usually included in the picture of the product details page), so as to facilitate price comparison to determine which e-commerce platform to purchase the target product, etc., which provides great convenience for users.

[0057] Combined with the above introduction, Figure 2 Schematic diagram of the interaction between the user, access service, proxy service and barrier-free service described in the present disclosure. Figure 2 As shown, the proxy service can obtain the first call request from the access service, and can obtain the first description information from the first call request. Then, the proxy service can call M accessibility services respectively. Assume that the value of M is 4. For the convenience of description, the four accessibility services are respectively called accessibility service 1, accessibility service 2, accessibility service 3 and accessibility service 4. The four accessibility services correspond to e-commerce platform 1, e-commerce platform 2, e-commerce platform 3 and e-commerce platform 4. In addition, the proxy service can obtain the search results returned by each accessibility service respectively. Assume that each accessibility service returns 15 search results respectively (the number of search results returned by different accessibility services in actual applications may be different), and each search result includes a merchant detail page image, a product detail page image and a comment detail page image. Then, the following processing can be performed for each accessibility service respectively: the low-quality search results are determined from the 15 search results returned by using the large model and the screening conditions and deleted. Assuming that the number of the remaining search results is 8, the 8 search results can be sorted, and the search results in the top 3 after sorting can be returned to the user. Among them, operations such as determining and deleting low-quality search results and sorting the remaining search results can be completed through the re-ranking service, and the corresponding functions of the re-ranking service can be integrated into the proxy service.

[0058] In addition, in some embodiments of the present disclosure, a third call request from the link identification service can also be obtained. The third call request carries the product link and target platform information. The target platform is the e-commerce platform corresponding to the product link identified by the link identification service after obtaining the product link from the access service. Thereafter, the accessibility service corresponding to the target platform can be called, and the predetermined image information of the product corresponding to the product link obtained by the accessibility service from the target platform can be obtained.

[0059] The user can access the access service through the HTTP protocol and provide a product link. The access service will call the link identification service, which will identify the e-commerce platform corresponding to the product link, such as e-commerce platform A, and use it as the target platform. The link identification service can then call the proxy service, such as sending a third call request to the proxy service, which may carry the product link and target platform information. The proxy service can further call the accessibility service corresponding to the target platform, and the accessibility service will copy the product link to the clipboard of the target platform. At this time, the target platform will automatically pop up a jump button. The accessibility service will jump to the product details page through the jump button and take a screenshot to obtain a product details page image. After that, the accessibility service can also access the merchant details page and the comment details page respectively, and can take screenshots respectively, so as to obtain the merchant details page image and the comment details page image respectively. Furthermore, the accessibility service can also return the acquired merchant details page image, product details page image and comment details page image to the proxy service, and then return them to the user.

[0060] The product link may be a product link shared by others to the user, etc. The user can view the returned merchant detail page images, product detail page images, and comment detail page images, so as to facilitate the user to intuitively and detailedly understand the product corresponding to the product link.

[0061] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are expressed as a series of action combinations, but those skilled in the art should be aware that the present disclosure is not limited by the described order of actions, because according to the present disclosure, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present disclosure.

[0062] The above is an introduction to the method embodiment. The following is a further explanation of the scheme disclosed in the present invention through an apparatus embodiment.

[0063] Figure 3 FIG. 3 is a schematic diagram of the structure of a retrieval processing device embodiment 300 based on a large model and barrier-free service according to the present disclosure. Figure 3As shown, it includes: a first processing module 301 and a second processing module 302.

[0064] The first processing module 301 is used to obtain first description information and call M accessibility services respectively, where M is a positive integer greater than 1. The first description information is description information that can determine the corresponding target product, and the target product is the product to be searched by the user.

[0065] The second processing module 302 is used to perform the following processing for any accessibility service: obtain the search results returned by the accessibility service, and use the big model to filter out the target search results from the search results, and return them to the user. The accessibility service is used to search in the corresponding e-commerce platform according to the first description information and determine the search results according to the principle of low price first.

[0066] By adopting the scheme described in the above-mentioned device embodiment, after obtaining the first description information, M accessibility services can be called respectively, so that the search results corresponding to the first description information can be retrieved from the corresponding e-commerce platforms using these M accessibility services, and the search results from each e-commerce platform can be screened using a large model and returned to the user, that is, the searches of each e-commerce platform are unified into one entrance, so that users can directly obtain the same product information on different e-commerce platforms, thereby facilitating users to compare prices, etc., thereby simplifying user operations and improving users' information acquisition efficiency. Moreover, by screening the search results, users' information acquisition efficiency is further improved and the quality of information obtained by users is improved.

[0067] In some embodiments of the present disclosure, the way in which the first processing module 301 obtains the first description information may include: obtaining a first call request from an access service, and obtaining first description information from the first call request, where the first description information is description information input by a user through an interactive interface.

[0068] Alternatively, in some embodiments of the present disclosure, the way in which the first processing module 301 obtains the first description information may include: obtaining a second call request from an access service, and obtaining second description information input by a user through an interactive interface from the second call request, the second description information being description information of a target product for which a corresponding description cannot be determined, calling M accessibility services respectively, and obtaining search results returned by each accessibility service for the second description information respectively, and returning them to the user, determining the product corresponding to the search result selected by the user from each search result as the target product, and determining the description information of the target product generated according to the selected search result as the first description information.

[0069] After obtaining the first description information, the first processing module 301 can call M accessibility services respectively. Accordingly, for any accessibility service, the second processing module 302 can perform the following processing respectively: obtain the search results returned by the accessibility service, and use the large model to filter out the target search results from the search results, and return them to the user.

[0070] In some embodiments of the present disclosure, for the search results corresponding to any accessibility service, the second processing module 302, in response to obtaining the filtering conditions set by the user for the target product, can use the large model to determine low-quality search results that do not meet the filtering conditions from the search results, and can delete the low-quality search results, and then can filter out the target search results from the remaining search results.

[0071] In some embodiments of the present disclosure, the screening condition may include N sub-conditions, where N is a positive integer. Accordingly, the second processing module 302 may use a large model to determine, for a search result corresponding to any accessibility service, low-quality search results that do not meet the screening condition from the search result, which may include: performing the following processing on any search result: using the large model to respectively determine the evaluation results of the search result corresponding to each sub-condition, and in response to determining that the search result does not meet any sub-condition based on the evaluation results, determining the search result as a low-quality search result.

[0072] Specifically, in some embodiments of the present disclosure, any search result may include: merchant details page pictures, product details page pictures, and review details page pictures. In addition, the number of large models may be N, and each sub-condition may correspond to one large model. Accordingly, the second processing module 302 may use the large model to determine, for any search result, the evaluation results of the search result corresponding to each sub-condition, including: for any sub-condition, performing the following processing: determining the picture corresponding to the sub-condition from the pictures included in the search result as the target picture, and constructing a prompt word based on the sub-condition, using the target picture and the prompt word as the input of the large model corresponding to the sub-condition, and obtaining the output evaluation result.

[0073] For any search result, after obtaining the evaluation results of the search result corresponding to each sub-condition, if it is determined according to the evaluation results that the search result does not meet any sub-condition, the second processing module 302 can determine the search result as a low-quality search result and delete the low-quality search result.

[0074] Afterwards, the second processing module 302 may filter out target search results from the remaining search results.

[0075] In some embodiments of the present disclosure, the N sub-conditions may include at least: a first sub-condition in which the number of negative reviews is less than a first threshold or the proportion of negative reviews is less than a second threshold, and the evaluation results corresponding to each sub-condition may include at least: a first evaluation result corresponding to the first sub-condition, the first evaluation result being the number of negative reviews or the proportion of negative reviews. Accordingly, the second processing module 302 may filter out target search results from the remaining search results in a manner that may include: sorting the remaining search results in ascending order of the values ​​of the first evaluation results, and in response to determining that the number of remaining search results is less than or equal to P, where P is a positive integer, determining all sorted search results as target search results, and in response to determining that the number of remaining search results is greater than P, determining the search results in the top P positions after sorting as target search results.

[0076] In addition, in some embodiments of the present disclosure, the method by which the second processing module 302 filters out the target search results from the remaining search results may also include: performing the following processing for any remaining search result: determining a comprehensive score of the search result in combination with the evaluation results of the search result corresponding to each sub-condition, sorting the remaining search results in order from large to small according to the values ​​of the comprehensive score, and in response to determining that the number of remaining search results is less than or equal to P, where P is a positive integer, determining all sorted search results as target search results; and in response to determining that the number of remaining search results is greater than P, determining the search results in the first P positions after sorting as target search results.

[0077] In addition, in some embodiments of the present disclosure, the first processing module 301 can also obtain a third call request from the link identification service, the third call request carries a product link and target platform information, the target platform is the e-commerce platform corresponding to the product link identified by the link identification service after obtaining the product link from the access service, and then the accessibility service corresponding to the target platform can be called, and the predetermined image information of the product corresponding to the product link obtained by the accessibility service from the target platform can be obtained.

[0078] Figure 3 The specific working process of the illustrated device embodiment can refer to the relevant description in the aforementioned method embodiment and will not be described in detail.

[0079] The solution disclosed in the present disclosure can be applied to the field of artificial intelligence, especially to the fields of large models, mobile Internet, computer vision and cloud computing. Artificial intelligence is a discipline that studies how computers can simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.). It has both hardware-level and software-level technologies. Artificial intelligence hardware technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, and big data processing. Artificial intelligence software technologies mainly include computer vision technology, speech recognition technology, natural language processing technology, machine learning / deep learning, big data processing technology, knowledge graph technology, and other major directions.

[0080] The description information and search results in the embodiments of this disclosure are not for a specific user and do not reflect the personal information of a specific user. In addition, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in the technical solution of this disclosure are in compliance with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0081] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0082] Figure 4 A schematic block diagram of an electronic device 400 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0083] like Figure 4 As shown, the device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 402 or a computer program loaded from a storage unit 408 to a 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 via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0084] A number of components in the device 400 are connected to the I / O interface 405, including: an input unit 406, such as a keyboard, a mouse, etc.; an output unit 407, such as various types of displays, speakers, etc.; a storage unit 408, such as a disk, an optical disk, etc.; and a communication unit 409, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 409 allows the device 400 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0085] The computing unit 401 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI, Artificial Intelligence) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSP, Digital Signal Processing), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 401 performs the various methods and processes described above, such as the methods described in the present disclosure. For example, in some embodiments, the methods described in the present disclosure may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 400 via the ROM 402 and / or the communication unit 409. When the computer program is loaded into the RAM 403 and executed by the computing unit 401, one or more steps of the methods described in the present disclosure may be performed. Alternatively, in other embodiments, the computing unit 401 may be configured to execute the method described in the present disclosure in any other appropriate manner (for example, by means of firmware).

[0086] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard parts (ASSPs), system on chip systems (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 that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0087] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0088] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory (EPROM, Electronically Programmable Read-Only Memory), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM, Compact Disc Read-Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0089] 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 cathode ray tube (CRT) or a liquid crystal display (LCD) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) through which the user can provide input to the computer. Other types 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 input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0090] 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 with a graphical user interface or a web browser through which a user can interact with implementations 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 by digital data communication (e.g., a communication network) in any form or medium. Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0091] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0092] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0093] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A retrieval processing method based on a large model and barrier-free service, comprising: Obtaining first description information, and calling M accessibility services respectively, where M is a positive integer greater than 1, and the first description information is description information capable of determining a corresponding target product, and the target product is a product to be searched by the user; For any accessibility service, the following processing is performed respectively: the search results returned by the accessibility service are obtained, and the target search results are filtered out from the search results using a large model, and returned to the user. The accessibility service is used to search the corresponding e-commerce platform according to the first description information and determine the search results according to the principle of low price first.

2. The method according to claim 1, wherein: The obtaining of the first description information includes: obtaining a first call request from an access service, and obtaining the first description information from the first call request, where the first description information is description information input by the user through an interactive interface.

3. The method according to claim 1, wherein: The obtaining of the first description information comprises: Obtaining a second call request from the access service, and obtaining second description information input by the user through the interactive interface from the second call request, where the second description information is description information of the target product for which a corresponding one cannot be determined; respectively calling the M accessibility services, and respectively obtaining search results returned by each accessibility service for the second description information, and returning the results to the user; The commodity corresponding to the search result selected by the user from among the search results is determined as the target commodity, and the description information of the target commodity generated according to the selected search result is determined as the first description information.

4. The method according to claim 1, 2 or 3, wherein: The step of using the large model to filter out target search results from the search results includes: In response to obtaining the screening condition set by the user for the target product, using the large model to determine low-quality search results that do not meet the screening condition from the search results; The low-quality search results are deleted, and the target search results are screened out from the remaining search results.

5. The method according to claim 4, wherein: The screening condition includes N sub-conditions, where N is a positive integer; The step of using the large model to determine low-quality search results that do not meet the screening conditions from the search results includes: For any search result, the following processing is performed respectively: the evaluation results of the search result corresponding to each sub-condition are determined using the large model respectively, and in response to determining that the search result does not meet any sub-condition based on the evaluation results, the search result is determined as the low-quality search result.

6. The method according to claim 5, wherein: Any search result includes: merchant details page pictures, product details page pictures and review details page pictures; The number of the large models is N, and each sub-condition corresponds to a large model; The step of using the large model to respectively determine the evaluation results of the search results corresponding to each sub-condition includes: For any sub-condition, the following processing is performed respectively: the picture corresponding to the sub-condition is determined from the pictures included in the search results as the target picture, and a prompt word is constructed according to the sub-condition, and the target picture and the prompt word are used as the input of the large model corresponding to the sub-condition to obtain the output evaluation result.

7. The method according to claim 5, wherein: The N sub-conditions include at least: a first sub-condition that the number of negative reviews is less than a first threshold or the proportion of negative reviews is less than a second threshold; The evaluation results corresponding to each sub-condition include at least: a first evaluation result corresponding to the first sub-condition, the first evaluation result being the number of negative reviews or the proportion of negative reviews; The method of filtering out the target search result from the remaining search results includes: sorting the remaining search results in ascending order of the values ​​of the first evaluation result; in response to determining that the number of remaining search results is less than or equal to P, where P is a positive integer, all sorted search results are determined as the target search results; in response to determining that the number of remaining search results is greater than P, the search results in the first P positions after sorting are determined as the target search results.

8. The method according to claim 5, wherein: The step of screening out the target search result from the remaining search results includes: For any remaining search results, the following processing is performed respectively: combining the evaluation results of the search results corresponding to each sub-condition to determine the comprehensive score of the search results; The remaining search results are sorted in descending order according to the values ​​of the comprehensive score. In response to determining that the number of remaining search results is less than or equal to P, where P is a positive integer, the sorted search results are all determined as the target search results. In response to determining that the number of remaining search results is greater than P, the search results in the first P positions after sorting are determined as the target search results.

9. The method according to claim 1, 2 or 3, further comprising: Obtaining a third call request from the link identification service, wherein the third call request carries a product link and target platform information, wherein the target platform is an e-commerce platform corresponding to the product link identified by the link identification service after obtaining the product link from the access service; The barrier-free service corresponding to the target platform is called, and predetermined image information of the product corresponding to the product link obtained by the barrier-free service from the target platform is obtained.

10. A retrieval processing device based on a large model and barrier-free service, comprising: A first processing module and a second processing module; The first processing module is used to obtain first description information and call M accessibility services respectively, where M is a positive integer greater than 1, and the first description information is description information capable of determining a corresponding target product, and the target product is a product to be searched by the user; The second processing module is used to perform the following processing for any accessibility service: obtain the search results returned by the accessibility service, and use the big model to filter out target search results from the search results, and return them to the user. The accessibility service is used to search in the corresponding e-commerce platform according to the first description information and determine the search results according to the principle of low price first.

11. The device according to claim 10, wherein: The first processing module obtains a first call request from an access service, and obtains the first description information from the first call request, where the first description information is description information input by the user through an interactive interface.

12. The device according to claim 10, wherein: The first processing module obtains a second call request from the access service, and obtains second description information input by the user through the interactive interface from the second call request, the second description information being description information of a target product for which a corresponding description cannot be determined, and calls the M accessibility services respectively, and obtains search results returned by each accessibility service for the second description information respectively, and returns them to the user, determines the product corresponding to the search result selected by the user from each search result as the target product, and determines the description information of the target product generated according to the selected search result as the first description information.

13. The device according to claim 10, 11 or 12, wherein: For any accessibility service, in response to obtaining the filtering conditions set by the user for the target product, the second processing module uses the large model to determine low-quality search results that do not meet the filtering conditions from the search results returned by the accessibility service, deletes the low-quality search results, and filters out the target search results from the remaining search results.

14. The device according to claim 13, wherein: The screening condition includes N sub-conditions, where N is a positive integer; The second processing module performs the following processing for any search result: using the large model to determine the evaluation results of the search result corresponding to each sub-condition, in response to determining that the search result does not meet any sub-condition based on the evaluation results, determining the search result as the low-quality search result.

15. The device according to claim 14, wherein: Any search result includes: merchant details page pictures, product details page pictures and review details page pictures; The number of the large models is N, and each sub-condition corresponds to a large model; The second processing module performs the following processing for any sub-condition: determines the picture corresponding to the sub-condition from the pictures included in the search results as the target picture, and constructs a prompt word based on the sub-condition, and uses the target picture and the prompt word as the input of the large model corresponding to the sub-condition to obtain the output evaluation result.

16. The device according to claim 14, wherein: The N sub-conditions include at least: a first sub-condition that the number of negative reviews is less than a first threshold or the proportion of negative reviews is less than a second threshold; The evaluation results corresponding to each sub-condition include at least: a first evaluation result corresponding to the first sub-condition, the first evaluation result being the number of negative reviews or the proportion of negative reviews; The second processing module sorts the remaining search results in ascending order according to the values ​​of the first evaluation results. In response to determining that the number of remaining search results is less than or equal to P, where P is a positive integer, the sorted search results are all determined as the target search results. In response to determining that the number of remaining search results is greater than P, the search results in the first P positions after sorting are determined as the target search results.

17. The device according to claim 14, wherein: The second processing module performs the following processing on any remaining search result: determines a comprehensive score of the search result in combination with the evaluation results of the search result corresponding to each sub-condition; sorts the remaining search results in descending order of the values ​​of the comprehensive score; in response to determining that the number of remaining search results is less than or equal to P, where P is a positive integer, determines all sorted search results as the target search results; in response to determining that the number of remaining search results is greater than P, determines the search results in the first P positions after sorting as the target search results.

18. The device according to claim 10, 11 or 12, wherein: The first processing module is further used to obtain a third call request from a link identification service, the third call request carrying a product link and target platform information, the target platform being the e-commerce platform corresponding to the product link identified by the link identification service after obtaining the product link from the access service, calling the accessibility service corresponding to the target platform, and obtaining the predetermined image information of the product corresponding to the product link obtained by the accessibility service from the target platform.

19. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed 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 perform the method according to any one of claims 1 to 9.

20. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to make a computer execute the method according to any one of claims 1 to 9.

21. A computer program product, comprising a computer program / instruction, wherein when the computer program / instruction is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

Images