Method and system for evaluating product recall indicator, electronic device and storage medium

Abstract

The application provides a commodity recall index evaluation method and system, an electronic device and a storage medium, comprising: pulling commodity search terms and associated pre-stored commodity aggregation information of commodities in a data warehouse to construct a commodity search index; receiving and extracting commodity search terms in a to-be-processed search term input by a query interface, querying the commodity search index based on a recall algorithm and using the extracted commodity search terms to obtain first recall commodities; screening the first recall commodities to determine target recall commodities; obtaining target relevance between the target recall commodities and the commodity search terms based on the target recall commodities, the commodity search terms and an optimal relevance model; and comparing the target relevance between the target recall commodities and the commodity search terms with pre-stored historical relevance to evaluate a commodity recall index. By establishing an offline index, the application range of the evaluation method is expanded, and the efficiency and accuracy of commodity recall are guaranteed. In addition, by shunting tests, the relevance model is optimized to improve the commodity recall index.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of electronic commerce, and in particular to a commodity recall index evaluation method and system, an electronic device and a storage medium. BACKGROUND

[0002] It is a common function of an e-commerce system to query corresponding related commodities through a search term. The accuracy of the searched commodities affects the user experience. However, the current e-commerce system uses a common weighting technique to calculate the relevance of the search term and the commodity, which has poor effect and lacks a solution for optimizing the relevance calculation. In addition, there is currently a lack of a method for evaluating commodity recall indexes in an offline scenario in the search field to facilitate the evaluation of the recall indexes at any time.

[0003] Therefore, there is an urgent need for a method for efficiently evaluating commodity recall indexes in an offline scenario to solve the above technical problems.

[0004] SUMMARY

[0005] To solve the problems of the prior art, the main purpose of the present application is to provide a commodity recall index evaluation method and system, an electronic device and a storage medium to solve the above technical problems of the prior art.

[0006] To achieve the above purpose, the first aspect of the present application provides a commodity recall index evaluation method:

[0007] Pulling commodity search terms and associated pre-stored commodity aggregation information in a data warehouse to construct a commodity search index, wherein the commodity search term corresponds to at least one item of commodity aggregation information;

[0008] Receiving and extracting commodity search terms in a to-be-processed search term input by a query interface, querying the commodity search index based on a recall algorithm and using the extracted commodity search terms to obtain first recall commodities;

[0009] Screening the first recall commodities to determine target recall commodities;

[0010] Based on the target recall commodities, the commodity search terms and the optimal relevance model, obtaining the target relevance of the target recall commodities and the commodity search terms;

[0011] Comparing the target relevance of the target recall commodities and the commodity search terms with pre-stored historical relevance to evaluate the commodity recall index.

[0012] In some embodiments, the method further comprises:

[0013] Extracting a pre-set number of target search terms, querying the commodity search index based on a recall algorithm and the target search terms to obtain test recall commodities;

[0014] If the number of test recall commodities recalled based on the target search term is less than a preset threshold, it is determined that the commodity search index is incomplete;

[0015] The commodity search index is constructed by re-pulling commodity aggregation information of the commodity search term and associated pre-stored commodities in the data warehouse;

[0016] The above steps are repeated until the number of test recall commodities recalled based on the target search term is greater than or equal to the preset threshold.

[0017] In some embodiments, the commodity search term in the to-be-processed search term is extracted, comprising:

[0018] The to-be-processed search term is subjected to word segmentation processing;

[0019] The commodity search term in the to-be-processed search term after the word segmentation processing is identified based on a natural language processing technology.

[0020] In some embodiments, the method further comprises:

[0021] Before obtaining the first recall commodity, the commodity search term is divided to set multiple groups of commodity search groups according to a user identifier of a historical user logging into a commodity system and / or a device identifier of a historical device accessing the commodity system;

[0022] A corresponding control correlation model is set for each group of commodity search groups, wherein parameters of the control correlation model corresponding to each group of commodity search groups are different;

[0023] Based on the multiple groups of divided commodity search groups and the corresponding control correlation models, a commodity recall indicator corresponding to the commodity search term in each group of commodity search groups is obtained.

[0024] In some embodiments, the method further comprises:

[0025] The recall rate and the precision rate of the commodity recall indicator corresponding to each group of commodity search terms are calculated;

[0026] The control correlation model corresponding to the commodity search term with the highest recall rate and the highest precision rate is determined as the optimal correlation model.

[0027] In some embodiments, the method further comprises:

[0028] The commodity search term is extracted based on a preset rule, and a pre-stored commodity constituting a commodity pool is selected;

[0029] The historical correlation between the pre-stored commodity and the commodity search term is set;

[0030] annotating the pre-stored goods and the historical relevance of the goods search term and establishing a relevance table of the pre-stored goods and the goods search term and storing to the data warehouse for pulling.

[0031] In some embodiments, the evaluating the goods recall indicator comprises calculating a recall rate and a precision rate, and the comparing the target relevance of the target recall goods and the goods search term with the pre-stored historical relevance of the pre-stored goods and the goods search term to evaluate the goods recall indicator comprises:

[0032] calculating the recall rate, wherein the recall rate is equal to a ratio of the number of the target recall goods corresponding to the goods search term to the number of the pre-stored goods corresponding to the goods search term in the data warehouse;

[0033] screening out duplicate goods of the target recall goods corresponding to the goods search term and the pre-stored goods in the data warehouse;

[0034] calculating the precision rate, wherein the precision rate is equal to a ratio of the number of the duplicate goods to the number of the target recall goods.

[0035] In a second aspect, the present application provides a goods recall indicator evaluation system, the system comprising:

[0036] an index processing module configured to pull goods aggregation information of a goods search term and associated pre-stored goods in a data warehouse to construct a goods search index, wherein the goods search term corresponds to at least one item of goods aggregation information;

[0037] a goods recall module configured to receive and extract a goods search term in a to-be-processed search term input by a query interface, query the goods search index based on a recall algorithm and using the extracted goods search term to obtain first recall goods;

[0038] a data screening module configured to screen the first recall goods to determine target recall goods;

[0039] a data analysis module configured to obtain a target relevance of the target recall goods and the goods search term based on the target recall goods, the goods search term and an optimal relevance model;

[0040] a data processing module configured to compare the target relevance of the target recall goods and the goods search term with pre-stored historical relevance to evaluate a goods recall indicator.

[0041] In a third aspect, the present application provides an electronic device, the electronic device comprising:

[0042] one or more processors;

[0043] and a memory associated with the one or more processors, the memory for storing program instructions that, when read and executed by the one or more processors, perform the following operations:

[0044] pulling commodity search terms and associated pre-stored commodity commodity aggregate information of commodities in a data warehouse to build a commodity search index, wherein the commodity search terms correspond to at least one commodity aggregate information;

[0045] receiving and extracting commodity search terms in a to-be-processed search term input by a query interface, querying the commodity search index based on a recall algorithm and using the extracted commodity search terms to obtain first recall commodities;

[0046] screening the first recall commodities to determine target recall commodities;

[0047] obtaining target relevance of the target recall commodities and the commodity search terms based on the target recall commodities, the commodity search terms, and an optimal relevance model;

[0048] comparing the target relevance of the target recall commodities and the commodity search terms with pre-stored historical relevance to evaluate a commodity recall index.

[0049] In a fourth aspect, the present application also provides a computer readable storage medium, the storage medium storing a computer program, the computer program causing a computer to perform the following operations:

[0050] pulling commodity search terms and associated pre-stored commodity commodity aggregate information of commodities in a data warehouse to build a commodity search index, wherein the commodity search terms correspond to at least one commodity aggregate information;

[0051] receiving and extracting commodity search terms in a to-be-processed search term input by a query interface, querying the commodity search index based on a recall algorithm and using the extracted commodity search terms to obtain first recall commodities;

[0052] screening the first recall commodities to determine target recall commodities;

[0053] obtaining target relevance of the target recall commodities and the commodity search terms based on the target recall commodities, the commodity search terms, and an optimal relevance model;

[0054] comparing the target relevance of the target recall commodities and the commodity search terms with pre-stored historical relevance to evaluate a commodity recall index.

[0055] The beneficial effects achieved by the present application are:

[0056] The application provides a commodity recall index evaluation method, comprising the following steps: pulling commodity search words and associated pre-stored commodity aggregation information of commodities in a data warehouse to construct a commodity search index, wherein the commodity search words correspond to at least one item of commodity aggregation information; receiving and extracting commodity search words in a to-be-processed search word input by a query interface; querying the commodity search index based on a recall algorithm and using the extracted commodity search words to obtain first recall commodities; screening the first recall commodities to determine target recall commodities; obtaining target correlation between the target recall commodities and the commodity search words based on the target recall commodities, the commodity search words and an optimal correlation model; and comparing the target correlation between the target recall commodities and the commodity search words with pre-stored historical correlation to evaluate a commodity recall index. By establishing an offline commodity search index, the application expands the application range of commodity recall index evaluation while ensuring the efficiency and accuracy of commodity recall. In addition, the application also proposes a shunting test, optimizes the correlation model and further improves the commodity recall index. BRIEF DESCRIPTION OF DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0058] Figure 1 FIG. 1 is a first schematic diagram of a commodity recall index evaluation method provided by the present application;

[0059] Figure 2 FIG. 2 is a second schematic diagram of a commodity recall index evaluation method provided by the present application;

[0060] Figure 3 FIG. 3 is a system architecture diagram of a commodity recall index evaluation system provided by the present application;

[0061] Figure 4 FIG. 4 is a structure diagram of an electronic device provided by the present application. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solutions and advantages of the present application more clear, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, but not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0063] It should be understood that in the description of the present application, unless the context clearly requires otherwise, the terms "comprise", "comprising", and the like are to be construed in an inclusive sense as opposed to an exclusive or exhaustive sense; that is to say, in the sense of "including, but not limited to".

[0064] It should also be understood that the terms "first", "second" and the like are used only for descriptive purposes and should not be construed as indicating or implying relative importance. In addition, in the description of the present application, unless otherwise stated, the meaning of "multiple" is two or more.

[0065] It should be noted that the terms "S1", "S2" and the like are only for the purpose of describing the steps and do not mean to indicate or imply the order or position, nor to limit the present application, which is only for the convenience of describing the method of the present application, and cannot be understood as indicating the order of the steps. In addition, the technical solutions of various embodiments can be combined with each other, but it must be based on the realization of ordinary skilled in the art, when the combination of technical solutions appears contradictory or unachievable, it should be considered that the combination of technical solutions does not exist, also not within the scope of protection required by the present application.

[0066] Embodiment one

[0067] Reference Figure 1 The present application provides a kind of commodity recall index evaluation method, using the method for evaluating the method of commodity recall index specifically includes:

[0068] S10, select and label the historical relevance of commodity search term and pre-stored commodity.

[0069] Specifically, based on the pre-set rule extraction commodity search term and select the pre-stored commodity that constitutes commodity pool. Preferably, using pre-set rule in extraction commodity search term can be according to the distribution ratio of high frequency, medium frequency and low frequency user request volume, hierarchical random extraction of different commodity search term;At the same time, in order to ensure that the extracted commodity search term can query specific commodity, when extracting commodity search term, operation words such as "all the special field of back force", "Olay big brand day" and "only card" should be eliminated. Preferably, based on the pre-set rule for selecting pre-stored commodity that constitutes commodity pool can include obtaining the number of exposures of commodity, classified according to the number of exposures: high exposure commodity, medium exposure commodity and low exposure commodity, a certain number of commodities are randomly extracted from the three types of commodities to constitute commodity pool. According to the rule setting and labeling the relationship between commodity search term and commodity, such as according to the category and brand of commodity, set the historical relevance of commodity and search term, that is, strong correlation, weak correlation and no correlation. The association table of commodity search term and pre-stored commodity after labeling historical relevance is stored in data warehouse, wherein the association table is constructed according to the strong correlation and weak correlation of commodity search term and pre-stored commodity.

[0070] S20. Call the query interface based on the product search terms to return the target relevance between the product search terms and the product.

[0071] Specifically, the process of obtaining target relevance using the query interface includes the following steps:

[0072] S21. Pull product search terms and associated pre-stored product aggregation information (SPU, Standard Product Unit) from the data warehouse to construct a product search index, wherein the product search terms correspond to at least one product aggregation information.

[0073] Preferably, this embodiment of the application is based on ETL (Extract-Transform-Load) technology, which retrieves product search terms and associated pre-stored product aggregation information from the data warehouse, and converts them into a format for standardization. To ensure the simplicity and accuracy of the constructed product search index, only the pre-stored product aggregation information is retrieved when data is pulled, and other labeled information, such as relevance, is not retrieved.

[0074] Preferably, to ensure the integrity of the extracted data during the data extraction process, this embodiment further proposes that after constructing the product search index, a preset number of target search terms are extracted from the product search index. Based on a recall algorithm, test products related to the target search terms are recalled from the product search index. If the number of recalled test products is less than a preset threshold, the product search index is determined to be incomplete. In this case, data, namely product search terms and product aggregation information, is retrieved from the data warehouse again to construct a new product search index. After constructing the new product search index, the above steps are repeated, and the test products recalled using the target search terms are used to determine the integrity of the newly built product search index. When the number of test products is greater than or equal to the preset threshold, the newly built product search index is determined to be complete. The preset number is defined by the user and can be 200, and the preset threshold is the number of target search terms that should normally be recalled based on this preset number. It is worth noting that subsequent operations on the product search index are all based on the verified integrity of the product search index.

[0075] S22. Obtain product search terms based on the search terms to be processed input from the query interface, and query the product search index based on the obtained product search terms to obtain the first recalled product.

[0076] Specifically, the search term to be processed can be subjected to text normalization processing, including special symbol filtering, case normalization, traditional Chinese to simplified Chinese conversion, and full-width to half-width conversion, to address different differences in inputting search terms by different users, and facilitate subsequent extraction of product search terms in the search term to be processed. The search term to be processed is subjected to word segmentation processing based on a preset word segmentation algorithm, which includes but is not limited to a dictionary-based word segmentation algorithm, an understanding-based word segmentation algorithm, and a statistical-based word segmentation algorithm. The product search term in the search term to be processed after word segmentation processing is identified by using natural language processing technology; the natural language processing technology is a technology that enables a computer to understand and accept human input instructions in natural language, and to complete translation from one language to another; the product search term in the search term to be processed is obtained by using analysis technologies such as morphological analysis, syntactic analysis, pragmatic analysis, contextual analysis, and natural language generation. The semantic recognition by using the natural language processing technology is a conventional technical means in the field, and will not be described herein.

[0077] Based on the product search term extracted from the search term to be processed and a recall algorithm, a first recalled product is determined by querying a product search index; preferably, the recall algorithm can be an ES (Elastic Search) recall algorithm, which is a distributed and highly scalable full-text retrieval search engine that can also provide near-real-time indexing, analysis, and search functions, and has a high recall accuracy.

[0078] S23, screening the first recalled product to determine a target recalled product.

[0079] Specifically, since the product search index is constructed based on SPU information in the embodiment, the first recalled product is only an SPU-level product; the SPU information is the smallest unit of product information aggregation, and is a collection of reusable and easily searchable standardized information; products with the same attribute values and characteristics can be referred to as an SPU, for example, in the case of a mobile phone-apple mobile phone-apple 6, the apple 6 is an SPU. The SPU-level product can also be screened based on SKU (Stock Keeping Unit) information; in the embodiment, the optimal product is screened from the first recalled product as the target product based on a preset strategy, which includes price priority, inventory priority, and color priority strategies, for example, according to the price priority strategy, the product with the lowest price in the first recalled product is screened as the target recalled product.

[0080] S24, obtaining a target relevance between the target recalled product and the product search term based on the target recalled product, the product search term, and an optimal relevance model.

[0081] The correlation model can be a Boolean model, a TF / IDF model, a spatial vector model, or other models for calculating data correlation. Determining the target relevance of the product search term and the target recall product by using the correlation model is a conventional technical means in the field, and thus the present embodiment will not be described herein.

[0082] S30, comparing the target relevance of the target recall product and the product search term with the pre-stored historical relevance to evaluate the product recall index.

[0083] Specifically, evaluating the product recall index includes calculating the recall rate and the precision rate. The recall rate is equal to the ratio of the number of target recall products corresponding to the product search term to the number of pre-stored products corresponding to the product search term in the data warehouse. The target recall product corresponding to the product search term is the relevant product (strongly relevant + weakly relevant) returned by the query interface. The repeated products are screened out from the target recall products corresponding to the product search term and the pre-stored products in the data warehouse. The repeated products are the relevant products returned by the query interface and belong to the pre-stored products associated with the same product search term. Then, the precision rate is calculated, which is equal to the ratio of the number of repeated products to the number of target recall products.

[0084] In addition, it is worth noting that the present embodiment optimizes the correlation model used to ensure the accuracy of the correlation calculation. The present embodiment discloses that before obtaining the first recall product, the product search term is divided into multiple product search groups according to the user identifier of the historical user logged into the product system and / or the device identifier of the historical device accessing the product system. Then, a corresponding control correlation model is set for each product search group. One product search group corresponds to one correlation model, and the correlation models corresponding to each product search group are different. Based on the multiple divided product search groups and the corresponding control correlation models, the above steps S2 and S3 are repeated to calculate the product recall index corresponding to different control correlation models, i.e., to calculate the precision rate and the recall rate corresponding to different correlation models. All control correlation models are screened to determine the control correlation model with the highest recall rate and the highest precision rate as the optimal correlation model.

[0085] Embodiment Two

[0086] Corresponding to the above-mentioned embodiment one, the present embodiment also provides a product recall index evaluation method, as shown in Figure 2 as follows:

[0087] 2100, pulling the product aggregation information of the product search term and the associated pre-stored product in the data warehouse to construct a product search index, wherein the product search term corresponds to at least one product aggregation information;

[0088] Preferably, the method further comprises:

[0089] 2110、extracting a preset number of target search terms, and querying the commodity search index based on a recall algorithm and the target search terms to obtain test recall commodities;

[0090] 2120、if the number of test recall commodities recalled based on the target search terms is less than a preset threshold, determining that the commodity search index is incomplete;

[0091] 2130, re-pull the commodity search terms and associated pre-stored commodity aggregation information in the data warehouse to construct the commodity search index;

[0092] 2140, repeat the above steps until the number of test recall commodities recalled based on the target search terms is greater than or equal to a preset threshold.

[0093] 2200, receiving and extracting commodity search terms in a to-be-processed search term input by a query interface, querying the commodity search index based on a recall algorithm and using the extracted commodity search terms to obtain first recall commodities;

[0094] Preferably, the extracted commodity search terms in the to-be-processed search term include:

[0095] 2210, performing word segmentation processing on the to-be-processed search term;

[0096] 2220, identifying commodity search terms in the to-be-processed search term after word segmentation processing based on natural language processing technology.

[0097] Preferably, the method further comprises:

[0098] 2230, before obtaining the first recall commodities, dividing the commodity search terms to set multiple groups of commodity search groups according to user identifiers of historical users logged into the commodity system and / or device identifiers of historical devices accessing the commodity system;

[0099] 2240, setting a corresponding control correlation model for each group of commodity search groups, wherein parameters of the control correlation model corresponding to each group of commodity search groups are different;

[0100] 2250, obtaining a commodity recall indicator corresponding to a commodity search term in each group of commodity search groups based on the multiple groups of divided commodity search groups and the corresponding control correlation models.

[0101] Preferably, the method further comprises:

[0102] 2260, calculating a recall rate and a precision rate in the commodity recall indicator corresponding to each group of commodity search terms;

[0103] 2270, determining the control correlation model corresponding to the commodity search term with the highest recall rate and the highest precision rate as the optimal correlation model.

[0104] 2300, screening the first recalled commodity to determine a target recalled commodity;

[0105] 2400, obtaining a target correlation between the target recalled commodity and the commodity search term based on the target recalled commodity, the commodity search term, and the optimal correlation model;

[0106] 2500, comparing the target correlation between the target recalled commodity and the commodity search term with a pre-stored historical correlation to evaluate a commodity recall indicator.

[0107] Preferably, the method further comprises:

[0108] 2510, extracting the commodity search term and selecting pre-stored commodities to form a commodity pool based on a pre-set rule;

[0109] 2520, setting a historical correlation between the pre-stored commodities and the commodity search term;

[0110] 2530, labeling the historical correlation between the pre-stored commodities and the commodity search term, establishing an association table between the pre-stored commodities and the commodity search term, and storing the association table in the data warehouse for pulling.

[0111] Embodiment three

[0112] The embodiments of the present application also provide a commodity recall indicator evaluation system, as shown in Figure 3 The system comprises:

[0113] An index processing module 310 is configured to pull commodity aggregation information of commodity search terms and associated pre-stored commodities in a data warehouse to construct a commodity search index, wherein the commodity search term corresponds to at least one item of commodity aggregation information;

[0114] A commodity recall module 320 is configured to receive and extract commodity search terms in a to-be-processed search term input by a query interface, query the commodity search index based on a recall algorithm and using the extracted commodity search terms to obtain first recalled commodities;

[0115] A data screening module 330 is configured to screen the first recalled commodities to determine a target recalled commodity;

[0116] A data analysis module 340 is configured to obtain a target correlation between the target recalled commodity and the commodity search term based on the target recalled commodity, the commodity search term, and an optimal correlation model;

[0117] The data processing module 350 is configured to compare the target relevance of the target recall commodity and the target search word with the pre-stored historical relevance to evaluate the commodity recall index.

[0118] In some embodiments, the system further comprises a data verification module 360 configured to extract a preset number of target search words, query the commodity search index based on a recall algorithm and the target search words to obtain test recall commodities, and determine that the commodity search index is incomplete if the number of test recall commodities recalled based on the target search words is less than a preset threshold. The index processing module 310 re-pulls commodity aggregation information of the commodity search words and associated pre-stored commodities in the data warehouse to construct the commodity search index. The above steps are repeated until the number of test recall commodities recalled based on the target search words is greater than or equal to the preset threshold.

[0119] In some embodiments, the commodity recall module 320 is further configured to perform word segmentation processing on the to-be-processed search words, and identify commodity search words in the word segmentation processed to-be-processed search words based on natural language processing technology.

[0120] In some embodiments, the system further comprises a model optimization module 370 configured to divide the commodity search words to set multiple commodity search groups according to user identifiers of historical users logged into the commodity system and / or device identifiers of historical devices accessing the commodity system before obtaining the first recall commodity. The model optimization module 370 is further configured to set a corresponding control relevance model for each commodity search group, wherein parameters of the control relevance models corresponding to each commodity search group are different. The model optimization module 370 is further configured to obtain commodity recall indexes corresponding to commodity search words in each commodity search group based on the divided multiple commodity search groups and the corresponding control relevance models.

[0121] In some embodiments, the model optimization module 370 is further configured to calculate a recall rate and a precision rate of each commodity search word in the commodity recall index corresponding to each commodity search group, and determine that a control relevance model corresponding to a commodity search word with the highest recall rate and the highest precision rate is an optimal relevance model.

[0122] In some embodiments, the system further comprises a data preparation module 380 configured to extract the commodity search words and select pre-stored commodities to form a commodity pool based on a preset rule. The data preparation module 380 is further configured to set historical relevance of the pre-stored commodities and the commodity search words. The data preparation module 380 is further configured to label the historical relevance of the pre-stored commodities and the commodity search words, establish an association table of the pre-stored commodities and the commodity search words, and store the association table to the data warehouse for pulling.

[0123] Embodiment Four

[0124] According to all the above embodiments, the present embodiment provides an electronic device, comprising:

[0125] one or more processors; and a memory associated with the one or more processors, the memory for storing program instructions that, when read and executed by the one or more processors, perform the following operations:

[0126] pulling commodity search terms and associated pre-stored commodity commodity aggregation information in the data warehouse to construct a commodity search index, wherein the commodity search term corresponds to at least one item of commodity aggregation information;

[0127] receiving and extracting commodity search terms in the search term input by the query interface, querying the commodity search index based on the recall algorithm and using the extracted commodity search terms to obtain first recall commodities;

[0128] screening the first recall commodities to determine target recall commodities;

[0129] based on the target recall commodities, the commodity search terms and the optimal correlation model, obtaining the target correlation of the target recall commodities and the commodity search terms;

[0130] comparing the target correlation of the target recall commodities and the commodity search terms with the pre-stored historical correlation to evaluate the commodity recall index.

[0131] wherein, Figure 4 The exemplary architecture of the electronic device is shown, which can specifically include a processor 410, a video display adapter 411, a disk drive 412, an input / output interface 413, a network interface 414, and a memory 420. The above-mentioned processor 410, video display adapter 411, disk drive 412, input / output interface 413, network interface 414, and memory 420 can be connected by bus 430 for communication.

[0132] The processor 410 can be implemented by a general-purpose CPU (Central Processing Unit, central processor), a microprocessor, an application specific integrated circuit (ASIC), or one or more integrated circuits, etc., for executing related programs to implement the technical solutions provided by the present application.

[0133] The memory 420 can be implemented in the form of a ROM (Read Only Memory), a RAM (Random Access Memory), a static storage device, a dynamic storage device, etc. The memory 420 can store an operating system 421 for controlling operations performed by the electronic device 400, a basic input / output system (BIOS) 422 for controlling low-level operations of the electronic device 400. In addition, a web browser 423, a data storage management system 424, an icon font processing system 425, etc. can also be stored. The icon font processing system 425 described above can be an application program that specifically implements the operations of the above steps in the embodiments of the present application. In summary, when the technical solutions provided by the present application are implemented by software or firmware, the relevant program codes are stored in the memory 420 and executed by the processor 410.

[0134] The input / output interface 413 is configured to connect input / output modules to realize information input and output. The input / output modules can be configured as components in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. The input devices can include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output devices can include a display, a speaker, a vibrator, an indicator light, etc.

[0135] The network interface 414 is configured to connect a communication module (not shown in the figure) to realize communication interaction between the device and other devices. The communication module can realize communication through wired means (such as USB, network cable, etc.) or through wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0136] The bus 430 includes a path for transmitting information between various components (such as the processor 410, the video display adapter 411, the disk drive 412, the input / output interface 413, the network interface 414, and the memory 420) of the device.

[0137] In addition, the electronic device 400 can also obtain information of the specific obtaining condition from the virtual resource object obtaining condition information database to be used for condition judgment, etc.

[0138] It should be noted that although the above device only shows the processor 410, the video display adapter 411, the disk drive 412, the input / output interface 413, the network interface 414, the memory 420, the bus 430, etc., in the specific implementation process, the device can also include other components necessary for normal execution. In addition, those skilled in the art can understand that the above device can also only contain components necessary for implementing the solutions of the present application, and does not necessarily contain all the components shown in the figure.

[0139] Embodiment Five

[0140] Corresponding to all the above embodiments, the embodiments of the present application also provide a computer readable storage medium, characterized in that it stores a computer program, which causes a computer to operate as follows:

[0141] Pulling the commodity search terms and the associated pre-stored commodity aggregate information of commodities in the data warehouse to construct a commodity search index, wherein the commodity search terms correspond to at least one item of commodity aggregate information;

[0142] Receiving and extracting commodity search terms in the search terms input by the query interface, querying the commodity search index based on the recall algorithm and using the extracted commodity search terms to obtain first recall commodities;

[0143] Screening the first recall commodities to determine target recall commodities;

[0144] Based on the target recall commodities, the commodity search terms and the optimal correlation model, obtaining the target correlation between the target recall commodities and the commodity search terms;

[0145] Comparing the target correlation between the target recall commodities and the commodity search terms with the pre-stored historical correlation to evaluate the commodity recall index.

[0146] From the above description of the embodiments, those skilled in the art can clearly understand that the present application can be realized by means of software and the necessary general hardware platform. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes a number of instructions for causing a computer device (which can be a personal computer, a cloud server, or a network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments of the present application.

[0147] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and relevant parts can be referred to the descriptions in the method embodiments. The systems and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0148] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for evaluating a product recall indicator, characterized by, The method comprises: pulling commodity search terms and associated commodity aggregation information of pre-stored commodities in a data warehouse to construct a commodity search index, wherein the commodity search terms correspond to at least one item of commodity aggregation information, and the commodity search terms are hierarchically randomly selected according to the distribution proportion of high-frequency, medium-frequency and low-frequency user request volume; receiving and extracting commodity search terms in a to-be-processed search term input by a query interface, querying the commodity search index based on a recall algorithm and using the extracted commodity search terms to obtain first recall commodities; screening the first recall commodities to determine target recall commodities; obtaining target relevance of the target recall commodities and the commodity search terms based on the target recall commodities, the commodity search terms and an optimal relevance model; comparing the target relevance of the target recall commodities and the commodity search terms with pre-stored historical relevance to evaluate commodity recall indicators; wherein the obtaining of the optimal relevance model comprises: before obtaining the first recall commodities, dividing the commodity search terms according to user identifiers of historical users logged into a commodity system and / or device identifiers of historical devices accessing the commodity system to set multiple groups of commodity search groups; setting a corresponding contrast relevance model for each group of commodity search groups, wherein parameters of the contrast relevance models corresponding to each group of commodity search groups are different; respectively obtaining commodity recall indicators corresponding to commodity search terms in each group of commodity search groups based on the divided multiple groups of commodity search groups and the corresponding contrast relevance models; calculating the recall rate and the precision rate of the commodity recall indicators corresponding to each group of commodity search terms; determining that the contrast relevance model corresponding to the commodity search terms with the highest recall rate and the highest precision rate is the optimal relevance model.

2. The method of claim 1, wherein, The method further comprises: extracting a preset number of target search terms, querying the commodity search index based on a recall algorithm and the target search terms to obtain test recall commodities; if the number of test recall commodities recalled based on the target search terms is less than a preset threshold, determining that the commodity search index is incomplete; re-pulling commodity aggregation information of pre-stored commodities associated with the commodity search terms in the data warehouse to construct a commodity search index; repeating the above steps until the number of test recall commodities recalled based on the target search terms is greater than or equal to the preset threshold.

3. The method of claim 2, wherein, The extraction of commodity search terms in the to-be-processed search term comprises: performing word segmentation processing on the to-be-processed search term; identifying commodity search terms in the to-be-processed search term after the word segmentation processing based on natural language processing technology.

4. The method of claim 3, wherein, The method further comprises: extracting the commodity search terms based on a preset rule and selecting pre-stored commodities to form a commodity pool; setting historical relevance of the pre-stored commodities and the commodity search terms; labeling the historical relevance of the pre-stored commodities and the commodity search terms, establishing an association table of the pre-stored commodities and the commodity search terms and storing them to the data warehouse for pulling.

5. The method according to any of claims 3-4, characterized by, The evaluation of the commodity recall indicators comprises calculating the recall rate and the precision rate, and the comparison of the relevance of the target recall commodities and the commodity search terms with the pre-stored relevance of commodities and commodity search terms to evaluate the commodity recall indicators comprises: calculate a recall rate, the recall rate being equal to a ratio of the target recall commodity quantity to a pre-stored commodity quantity corresponding to the commodity search word in the data warehouse; screen out duplicate commodities from the target recall commodities corresponding to the commodity search word and the pre-stored commodities in the data warehouse; calculate a precision rate, the precision rate being equal to a ratio of the duplicate commodity quantity to the target recall commodity quantity.

6. A product recall indicator evaluation system characterized by, The system comprises: an index processing module configured to pull commodity aggregation information of commodity search words and associated pre-stored commodities in a data warehouse to construct a commodity search index, wherein the commodity search word corresponds to at least one item of commodity aggregation information, and the commodity search word can be randomly selected according to a distribution ratio of high-frequency, medium-frequency and low-frequency user request quantities; a commodity recall module configured to receive and extract a commodity search word in a to-be-processed search word input by a query interface, query the commodity search index based on a recall algorithm and using the extracted commodity search word to obtain first recall commodities; a data screening module configured to screen the first recall commodities to determine target recall commodities; a data analysis module configured to obtain target relevance of the target recall commodities to the commodity search word based on the target recall commodities, the commodity search word and an optimal relevance model; a data processing module configured to compare the target relevance of the target recall commodities to the commodity search word with pre-stored historical relevance to evaluate a commodity recall index; a model optimization module configured to: before obtaining the first recall commodities, divide the commodity search words according to user identifiers of historical users logging into a commodity system and / or device identifiers of historical devices accessing the commodity system to set multiple groups of commodity search groups; set a contrast relevance model corresponding to each group of commodity search groups, wherein parameters of the contrast relevance models corresponding to the commodity search groups are different; obtain a commodity recall index corresponding to a commodity search word in each group of commodity search groups based on the divided multiple groups of commodity search groups and the corresponding contrast relevance models; calculate a recall rate and a precision rate in the commodity recall index corresponding to each group of commodity search words; determine a contrast relevance model corresponding to the commodity search word with the highest recall rate and the highest precision rate as an optimal relevance model.

7. An electronic device, comprising: The electronic device comprises: one or more processors; and a memory associated with the one or more processors, the memory being configured to store program instructions, the program instructions being configured to perform the method of any one of claims 1-5 when read and executed by the one or more processors.

8. A computer-readable storage medium, characterized in that, The computer program causes the computer to perform the method of any one of claims 1-5. The computer program causes the computer to perform the method of any one of claims 1-5.

Images