An information retrieval result processing method, device, server, and storage medium

Abstract

The present disclosure relates to a method, apparatus, server, and storage medium for processing information retrieval results. The method may include: obtaining a candidate object set according to a search term, and determining a target topic set corresponding to the search term; initializing an information retrieval result queue; repeatedly performing the following steps until the candidate object set is an empty set: calculating a diversity index of each candidate object according to the object topics included in each candidate object in the candidate object set and the target topic set; determining the candidate objects whose diversity index meets a preset condition as retrieval result objects; adding the retrieval result objects to the tail of the information retrieval result queue, and deleting the retrieval result objects from the candidate object set; and if the candidate object set is an empty set, outputting the information retrieval result queue. The present disclosure can improve the topic coverage rate of retrieval results and enhance the diversity of retrieval results.

Description

Technical Field

[0001] The present disclosure relates to the technical field of information retrieval, and particularly to a method, apparatus, server, and storage medium for processing information retrieval results. Background Art

[0002] Information retrieval evaluation is an activity for evaluating the performance of an information retrieval system (mainly the ability to meet users' information needs). Through evaluation, the advantages and disadvantages of different information retrieval technologies and the impact of different factors on the information retrieval system can be evaluated, thereby promoting the continuous improvement of the research level in this field. The goal of an information retrieval system is to return accurate results as quickly and comprehensively as possible with less resource consumption. Among the evaluation indicators of an information retrieval system, there is a diversity evaluation indicator, which is used to characterize the diversity of the returned information retrieval results.

[0003] In related technologies, generally a series of retrieval results are obtained according to the search terms submitted by users, and after relevance ranking, they are returned to users; if only relevance ranking is performed, there will be problems such as excessive redundancy of retrieval results and inability to effectively cover users' intentions, resulting in weak diversity of retrieval results, inaccurate retrieval results, and thus poor user experience. Therefore, a method capable of improving the accuracy of information retrieval results needs to be proposed. Summary of the Invention

[0004] The present disclosure provides a method, apparatus, server, and storage medium for processing information retrieval results to at least solve the problem of inaccurate information retrieval results in related technologies. The technical solution of the present disclosure is as follows:

[0005] According to the first aspect of the embodiments of the present disclosure, a method for processing information retrieval results is provided, including:

[0006] Obtain an information retrieval request, where the information retrieval request includes search terms;

[0007] Obtain a candidate object set according to the search terms, and determine a target topic set corresponding to the search terms;

[0008] Initialize an information retrieval result queue;

[0009] Repeat the following steps until the candidate object set is an empty set:

[0010] Calculate the diversity index of each candidate object according to the object topic included in each candidate object in the candidate object set and the target topic set;

[0011] Determine the candidate object whose diversity index meets the preset condition as a retrieval result object;

[0012] Add the retrieved result object to the tail of the information retrieval result queue, and delete the retrieved result object from the candidate object set;

[0013] If the candidate object set is an empty set, output the information retrieval result queue.

[0014] In an exemplary embodiment, calculating the diversity index of each candidate object according to the object topics included in each candidate object in the candidate object set and the target topic set includes:

[0015] For each object topic included in each candidate object, match each object topic with each target topic in the target topic set one by one to obtain a matching topic corresponding to each candidate object;

[0016] Calculate the diversity index sub-item corresponding to the matching topic;

[0017] Based on the diversity index sub-items, obtain the diversity index of each candidate object.

[0018] In an exemplary embodiment, after determining the target topic set corresponding to the search term, it further includes:

[0019] Generate a first word vector corresponding to each target topic in the target topic set;

[0020] After obtaining the candidate object set according to the search term, it further includes:

[0021] Determine at least one object topic included in each candidate object in the candidate object set;

[0022] Generate a second word vector corresponding to each object topic.

[0023] In an exemplary embodiment, for each object topic included in each candidate object, matching each object topic with each target topic in the target topic set one by one to obtain a matching topic corresponding to each candidate object includes:

[0024] Determine whether each object topic included in each candidate object is a matching topic;

[0025] Based on the judgment result of each object topic of each candidate object, obtain a matching topic corresponding to each candidate object.

[0026] In an exemplary embodiment, determining whether each object topic included in each candidate object is a matching topic includes:

[0027] Calculate the distance between the second word vector corresponding to each object topic and the first word vector corresponding to each target topic in the target topic set one by one, to obtain a word vector distance set;

[0028] Compare the smallest word vector distance in the word vector set with a preset threshold;

[0029] When the smallest word vector distance is less than the preset threshold, determine each of the object topics as a matching topic.

[0030] In an exemplary embodiment, the calculating the diversity index sub-item corresponding to the matching topic includes:

[0031] Determine the number of candidate objects including the matching topic in the information retrieval result queue;

[0032] Based on a preset diversity index sub-item calculation formula and the number of candidate objects including the matching topic in the information retrieval result queue, calculate the diversity index sub-item corresponding to the matching topic.

[0033] In an exemplary embodiment, before obtaining the candidate object set according to the search term, it further includes:

[0034] Perform text recognition and image recognition on the content of each object in the information resource library, and determine the label information of each object according to the recognition result;

[0035] The obtaining the candidate object set according to the search term includes:

[0036] Segment the search term to obtain at least one search keyword;

[0037] Respectively determine a search object set corresponding to each search keyword from the information resource library; wherein each search keyword has a correlation with the label information of each search object in the corresponding search object set;

[0038] Generate the candidate object set based on the search object set.

[0039] In an exemplary embodiment, the determining the target topic set corresponding to the search term includes:

[0040] Analyze the search term to determine the potential search intent corresponding to the search term;

[0041] Perform topic mining based on the potential search intent corresponding to the search term to generate multiple target topics corresponding to the search term;

[0042] Generate the target topic set based on the multiple target topics corresponding to the search term.

[0043] In an exemplary embodiment, the candidate object whose diversity index meets the preset condition is the candidate object with the largest diversity index;

[0044] Determining the candidate object whose diversity index meets the preset condition as the retrieval result object includes:

[0045] When the number of candidate objects with the largest diversity index is greater than one, determine the number of non-matching topics included in each candidate object with the largest diversity index;

[0046] Determine the candidate object with the smallest number of non-matching topics as the retrieval result object.

[0047] According to a second aspect of the embodiments of the present disclosure, there is provided an information retrieval result processing device, including:

[0048] An information retrieval request acquisition unit, configured to execute acquiring an information retrieval request, where the information retrieval request includes a search term;

[0049] An information retrieval request processing unit, configured to execute obtaining a candidate object set according to the search term, and determining a target topic set corresponding to the search term;

[0050] A queue initialization unit, configured to execute initializing an information retrieval result queue;

[0051] A repeated execution unit, configured to execute repeatedly calling the following units until the candidate object set is an empty set:

[0052] A diversity index calculation unit, configured to execute calculating the diversity index of each candidate object according to the object topic included in each candidate object in the candidate object set and the target topic set;

[0053] A retrieval result object determination unit, configured to execute determining the candidate object whose diversity index meets the preset condition as the retrieval result object;

[0054] An update unit, configured to execute adding the retrieval result object to the tail of the information retrieval result queue, and deleting the retrieval result object from the candidate object set;

[0055] An information retrieval result generation unit, configured to execute outputting the information retrieval result queue if the candidate object set is an empty set.

[0056] In an exemplary embodiment, the diversity index calculation unit includes:

[0057] A theme matching unit, configured to perform, for each object theme included in each candidate object, matching each object theme with each target theme in the target theme set one by one to obtain a matching theme corresponding to each candidate object;

[0058] A diversity metric sub-item calculation unit, configured to perform calculating a diversity metric sub-item corresponding to the matching theme;

[0059] A diversity metric determination unit, configured to perform obtaining the diversity metric of each candidate object based on the diversity metric sub-item.

[0060] In an exemplary embodiment, the apparatus further includes:

[0061] A first word vector generation unit, configured to perform generating a first word vector corresponding to each target theme in the target theme set;

[0062] An object theme determination unit, configured to perform determining at least one object theme included in each candidate object in the candidate object set;

[0063] A second word vector generation unit, configured to perform generating a second word vector corresponding to each object theme.

[0064] In an exemplary embodiment, the theme matching unit includes:

[0065] A matching theme judgment unit, configured to perform judging whether each object theme included in each candidate object is a matching theme;

[0066] A matching result determination unit, configured to perform obtaining a matching theme corresponding to each candidate object based on the judgment result of each object theme of each candidate object.

[0067] In an exemplary embodiment, the matching theme judgment unit includes:

[0068] A word vector distance calculation unit, configured to perform calculating the distance between the second word vector corresponding to each object theme and the first word vector corresponding to each target theme in the target theme set one by one to obtain a word vector distance set;

[0069] A comparison unit, configured to perform comparing the smallest word vector distance in the word vector distance set with a preset threshold;

[0070] A matching theme determination unit, configured to perform determining each object theme as a matching theme when the smallest word vector distance is less than the preset threshold.

[0071] In an exemplary embodiment, the diversity metric sub-item calculation unit includes:

[0072] A first determination unit configured to determine the number of candidate objects in the information retrieval result queue that include the matching topic;

[0073] A first calculation unit configured to calculate a diversity metric sub-item corresponding to the matching topic based on a preset diversity metric sub-item calculation formula and the number of candidate objects in the information retrieval result queue that include the matching topic.

[0074] In an exemplary embodiment, the apparatus further includes:

[0075] A content recognition unit configured to perform text recognition and image recognition on the content of each object in the information resource library, and determine the tag information included in each object according to the recognition result;

[0076] The information retrieval request processing unit includes:

[0077] A search term word segmentation unit configured to perform word segmentation on the search term to obtain at least one search keyword;

[0078] A search object set determination unit configured to respectively determine a search object set corresponding to each search keyword from the information resource library; wherein each search keyword has a relevance to the tag information of each search object in the corresponding search object set;

[0079] A candidate object set determination unit configured to generate the candidate object set based on the search object set.

[0080] In an exemplary embodiment, the information retrieval request processing unit includes:

[0081] A search term analysis unit configured to perform analysis on the search term to determine a potential search intent corresponding to the search term;

[0082] A topic mining unit configured to perform topic mining based on the potential search intent corresponding to the search term to generate a plurality of target topics corresponding to the search term;

[0083] A target topic set generation unit configured to generate the target topic set based on the plurality of target topics corresponding to the search term.

[0084] In an exemplary embodiment, the candidate object for which the diversity metric meets the preset condition is the candidate object with the maximum diversity metric;

[0085] The retrieval result object determination unit includes:

[0086] a non-matching topic number determining unit configured to determine the number of non-matching topics contained in each candidate object with the largest diversity index when the number of candidate objects with the largest diversity index is greater than one;

[0087] The second determining unit is configured to determine the candidate object with the smallest number of non-matching topics as the retrieval result object.

[0088] According to a third aspect of an embodiment of the present disclosure, a server is provided, comprising: a processor; and a memory for storing instructions executable by the processor; wherein the processor is configured to execute the instructions to implement the information retrieval result processing method as described above.

[0089] According to a fourth aspect of an embodiment of the present disclosure, a storage medium is provided. When instructions in the storage medium are executed by a processor of a server, the server is enabled to execute the information retrieval result processing method as described above.

[0090] According to the fifth aspect of an embodiment of the present disclosure, a computer program product is provided, which includes a computer program, and the computer program is stored in a readable storage medium. At least one processor of a computer device reads and executes the computer program from the readable storage medium, so that the device executes the above-mentioned information retrieval result processing method.

[0091] The technical solutions provided by the embodiments of the present disclosure bring at least the following beneficial effects:

[0092] The present disclosure obtains an information retrieval request including a search term, obtains a candidate object set based on the search term, and determines a target object set corresponding to the search term; based on the calculation result of the diversity index of each candidate object in the candidate object set, selects candidate objects whose diversity index meets preset conditions from the candidate object set and adds them to the end of the information retrieval result queue in sequence; when the candidate object set is an empty set, determines the information retrieval result queue currently containing multiple target objects as the information retrieval result corresponding to the search term and outputs it. The present disclosure can ensure that the top-ranked candidate objects in the returned information retrieval results contain as many different topics as possible, thereby improving the topic coverage of the retrieval results, thereby effectively covering the user's intent, reducing the redundancy of the retrieval results, enhancing the diversity of the retrieval results, and improving the accuracy of the information retrieval results.

[0093] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] The accompanying drawings here are incorporated into the description and form a part of this description, showing embodiments consistent with the present disclosure, and are used together with the description to explain the principles of the present disclosure, and do not constitute an undue limitation of the present disclosure.

[0095] Figure 1 It is a schematic diagram of an application scenario shown according to an exemplary embodiment.

[0096] Figure 2 It is a flowchart of a method for processing information retrieval results shown according to an exemplary embodiment.

[0097] Figure 3 It is a flowchart of a method for generating a set of candidate objects shown according to an exemplary embodiment.

[0098] Figure 4 It is a flowchart of a method for generating a set of target topics shown according to an exemplary embodiment.

[0099] Figure 5 It is a flowchart of a method for calculating a diversity index of candidate objects shown according to an exemplary embodiment.

[0100] Figure 6 It is a flowchart of a method for judging matching topics shown according to an exemplary embodiment.

[0101] Figure 7 It is a flowchart of a method for determining matching topics shown according to an exemplary embodiment.

[0102] Figure 8 It is a flowchart of a method for calculating sub-items of a diversity index shown according to an exemplary embodiment.

[0103] Figure 9 It is a flowchart of a method for determining retrieval result objects shown according to an exemplary embodiment.

[0104] Figure 10 It is a schematic diagram of a device for processing information retrieval results shown according to an exemplary embodiment.

[0105] Figure 11 It is a schematic diagram of a server structure shown according to an exemplary embodiment. Detailed implementation manners

[0106] To enable those of ordinary skill in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0107] It should be noted that the terms "first", "second", etc. in the specification, claims and above-mentioned drawings of the present disclosure are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present disclosure described here can be implemented in an order other than those illustrated or described here. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. On the contrary, they are only examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0108] Please refer to Figure 1 , which shows a schematic diagram of the implementation environment provided by the embodiments of the present disclosure. The implementation environment may include: at least one first terminal 110 and a second terminal 120. At least one first terminal 110 and the second terminal 120 can perform data communication through a network.

[0109] Specifically, the first terminal 110 generates a corresponding information retrieval request according to the retrieval requirement, and sends the information retrieval request to the second terminal 120, where the information retrieval request includes a search term; after receiving the information retrieval request, the second terminal 120 searches in the information resource library according to the search term included in the information retrieval request, and obtains a set of candidate objects relevant to the search term; after the second terminal 120 processes the set of candidate objects using a corresponding processing strategy, it obtains an information retrieval result and returns it to the first terminal 110.

[0110] The first terminal 110 can communicate with the second terminal 120 based on the browser / server mode (B / S) or the client / server mode (C / S). The first terminal 110 may include: entity devices of types such as smart phones, tablet computers, laptop computers, digital assistants, smart wearable devices, vehicle-mounted terminals, servers, etc., and may also include software running on the entity devices, such as application programs, etc. The operating systems running on the first terminal 110 in the embodiments of the present disclosure may include, but are not limited to, Android system, IOS system, linux, windows, etc.

[0111] The second terminal 120 and the first terminal 110 can establish a communication connection through wire or wireless. The second terminal 120 may include an independently operating server, or a distributed server, or a server cluster composed of multiple servers, where the server may be a cloud server.

[0112] To solve the problem of inaccurate information retrieval results in the related art, the embodiments of the present disclosure provide a method for processing information retrieval results, and its execution subject may be Figure 1The second terminal in Figure 2 may specifically be a server for performing an information retrieval operation. Please refer to

[0113] S210. Obtain an information retrieval request, where the information retrieval request includes a search term.

[0114] The information retrieval request in this disclosure may be sent by a user according to actual retrieval needs. The information retrieval request may be triggered after the user enters a search term on a relevant search page. Here, entering the search term may be manual input or voice input; it may also be triggered after the user clicks on a corresponding recommended search term label. The embodiments of this disclosure do not make specific limitations.

[0115] S220. Obtain a candidate object set according to the search term, and determine a target topic set corresponding to the search term.

[0116] During the information retrieval process, the search term entered by the user can be used to search in the information resource library, and a candidate object set relevant to the search term is returned. The objects in this disclosure may specifically refer to documents, web pages, images, videos, etc. Among them, when specifically determining the candidate object set relevant to the search term, it can be determined based on the label information of each object in the information resource library. Specifically, the content of each object in the information resource library can be subjected to text recognition and image recognition, and the label information of each object can be determined according to the recognition results. For example, when the text content of a certain object contains words such as "apple", "red delicious apple", "Chinese pear-leaved crabapple", "flowering crabapple", etc., and / or images of "apple", "red delicious apple", "Chinese pear-leaved crabapple", "flowering crabapple", etc. are recognized from the image content contained in the object, the label information of the object can be determined to include "apple", "red delicious apple", "Chinese pear-leaved crabapple", "flowering crabapple", etc.

[0117] For the generation method of the candidate object set relevant to the search term, please refer to Figure 3 and includes:

[0118] S310. Segment the search term to obtain at least one search keyword.

[0119] S320. Respectively determine a search object set corresponding to the search keyword from the information resource library; where each search keyword is relevant to the label information of each search object in the corresponding search object set.

[0120] S330. Generate the candidate object set based on the search object set.

[0121] The reason for tokenizing the search term is that the search term may not be a single word and may be a phrase composed of multiple words. Therefore, it is first necessary to tokenize the search term so that the obtained search keys can be one or more. When the number of search keywords is one, the search keyword can be matched with the label information of each object in the information resource library to determine the label information relevant to the search keyword. Thus, the object corresponding to the relevant label information can be determined as the search object, and then a search object set can be generated. The search object set can be directly determined as the candidate object set. When the number of search keywords is greater than one, each search keyword can be respectively matched with the label information of each object in the information resource library to determine the label information relevant to each search keyword. Thus, the search object corresponding to each search term can be determined, and a search object set corresponding to each search keyword can be generated. Then, the intersection of the search object sets corresponding to each search keyword is obtained, and thus the candidate object set can be obtained.

[0122] Among them, in the process of determining the relevance between the search keyword and the label information of the search object, it can be checked whether the label information of each object in the resource information library contains the search keyword, or whether it contains synonyms or near-synonyms of the search keyword, etc., so as to determine whether the search keyword and the label information are relevant. In addition, in addition to the above method of determining relevance through the meaning of the search keyword and the meaning of the words contained in the label information, any method that can determine relevance based on the search keyword and the label information can be applied to this embodiment. Taking the above example for illustration, when the search keyword is "apple", the label information of some objects in the information resource library includes "apple", and the label information of some objects includes "red delicious apple", then these objects can all be determined as the search objects corresponding to the search keyword, and thus a search object set is generated.

[0123] Each object in the candidate object set is a candidate object relevant to the search term. Determining the candidate object set from the resource library through the search term can preliminarily screen the objects related to the search term. Thus, subsequent operations can all be based on the screened candidate object set, which can narrow the scope of object processing and improve processing efficiency. For the generation of the target topic set, it is obtained based on the analysis of the search term. The target topic can be used to describe the information of the search term in different categories, different fields, etc. For details, please refer to Figure 4 , which shows a method for generating a target topic set, and may include:

[0124] S410. Analyze the search term to determine the potential search intent corresponding to the search term.

[0125] S420. Perform topic mining based on the potential search intents corresponding to the search term, and generate multiple target topics corresponding to the search term.

[0126] S430. Generate the target topic set based on the multiple target topics corresponding to the search term.

[0127] For a search term, it may contain multiple aspects of meanings, and for each aspect, there are different search results. Therefore, it is necessary to comprehensively analyze the search term to determine the potential search intent corresponding to the search term. Here, the potential search intent can specifically be the search intents for multiple different aspects corresponding to the search term; perform topic mining on the potential search intents for each aspect respectively. During the process of topic mining, include as many classifications involved in the potential search intent of each aspect as possible and generate corresponding topics, so as to facilitate subsequent matching between the target topics generated by mining and the object topics of candidate objects, and generate diverse retrieval results.

[0128] Taking the search term "apple" as an example, the word "apple" can include at least three aspects of meanings. On the one hand, it is the apple as a fruit, on the other hand, it is the apple as an electronic device brand, and on the other hand, it is the apple as a song name; thus, the meanings of these three aspects can be regarded as the potential search intents corresponding to the search term "apple", and relevant resources can be searched at least from these three aspects. Further, for the potential search intent of each aspect, there can also be different classifications; for the potential search intent of the apple as a fruit, the classifications it involves can include: apple varieties, planting methods, nutritional values, effects, recipes, etc.; for the potential intent of the apple as an electronic device brand, the classifications it involves can include: apple mobile phones, apple computers, models, tutorials, evaluations, repairs, etc.; for the potential intent of the apple as a song name, the classifications it involves can include: song introductions, song videos, singing skills, user evaluations, etc. The classifications involved in each potential search intent can be regarded as the target topics corresponding to the potential search intent, and thus the target topic set corresponding to the search term "apple" can be generated.

[0129] S230. Initialize the information retrieval result queue.

[0130] Initialize the information retrieval result queue. The initialized information retrieval result queue can be an empty queue.

[0131] S240. Determine whether the candidate object set is an empty set.

[0132] S250. If the candidate object set is not an empty set, calculate the diversity index of each candidate object according to the object topics included in each candidate object in the candidate object set and the target topic set.

[0133] The diversity metrics in the embodiments of the present disclosure can be used to characterize the topic coverage of each current candidate object. For details, please refer to Figure 5 , which shows a method for calculating the diversity metrics of candidate objects. The method may include:

[0134] S510. For each object topic included in each candidate object, match each object topic with each target topic in the target topic set one by one to obtain the matching topics corresponding to each candidate object.

[0135] S520. Calculate the sub-items of the diversity metrics corresponding to the matching topics.

[0136] S530. Based on the sub-items of the diversity metrics, obtain the diversity metrics of each candidate object.

[0137] For the calculation of the diversity metrics of each candidate object, it is necessary to perform based on the object topics included in the candidate object. Specifically, it is also necessary to obtain the matching topics corresponding to each candidate object according to the object topics and target topics included in each candidate object; the diversity metrics of each candidate object are obtained by calculating the sub-items of the diversity metrics corresponding to the matching topics. The diversity metrics of each candidate object in the present disclosure can be used to describe the topic coverage rate of the corresponding candidate object. Among them, the diversity metrics of each candidate object can be directly calculated through relevant formulas, and the diversity metrics of each candidate object can be obtained quickly and conveniently, thereby improving the convenience of reflecting the topic coverage rate of the candidate object.

[0138] Among them, specifically through Figure 6 a shown method for judging matching topics to obtain the matching topics corresponding to each candidate object. The method may include:

[0139] S610. Judge whether each object topic included in each candidate object is a matching topic.

[0140] S620. Based on the judgment results of each object topic of each candidate object, obtain the matching topics corresponding to each candidate object.

[0141] To determine whether each object topic is a matching topic, it can be achieved by checking whether there is a target topic in the target topic set that matches the object topic. Since the target topic set generally includes multiple target topics, when specifically performing topic matching, for each candidate object, each object topic included in the candidate object needs to be respectively matched with multiple target topics to determine whether each object topic is a matching topic; and the object topic that matches the target topic is determined as the matching topic, so that the matching topic corresponding to each candidate object can be obtained. By matching each object topic included in each candidate object with each target topic one by one and correspondingly determining whether the object topic is a matching topic, it is possible to avoid missing topic matching, thereby improving the comprehensiveness of topic matching.

[0142] In the specific process of topic matching, different topics may be different in literal expression but have similar meanings. Therefore, when determining whether any two topics match or have similar meanings, the method of word vectors can be used to achieve this. By calculating the word vector distance between the two topics, it can be determined whether the two topics match or have similar meanings. Before performing topic matching, it is also necessary to generate the first word vector corresponding to each target topic in the target topic set, and determine at least one object topic included in each candidate object in the candidate object set, and generate the second word vector corresponding to each object topic. Among them, for the determination of the object topic of the candidate object, a method similar to the determination of label information can be used, that is, the object topics included in each candidate object are determined by methods such as character recognition and image recognition; and the object topics of each candidate object can be determined in advance, so as to facilitate direct acquisition and use during topic matching.

[0143] Specifically, please refer to Figure 7 , which shows a method for determining a matching topic, and the method may include:

[0144] S710. Calculate the distance between the second word vector corresponding to each object topic and the first word vector corresponding to each target topic in the target topic set one by one, and obtain a set of word vector distances.

[0145] S720. Compare the smallest word vector distance in the word vector set with a preset threshold.

[0146] S730. When the smallest word vector distance is less than the preset threshold, determine each object topic as a matching topic.

[0147] Specifically, the word vector of the target topic in the target topic set can be expressed as:

[0148]

[0149] Based on the word vectors corresponding to each target topic, a word vector sequence corresponding to the target topic set can be generated, which can be expressed as:

[0150] Vector query = wordembedding(topic query ) (2)

[0151] For each candidate object, the word vector of the object topic included can be expressed as:

[0152]

[0153] Calculate the vector distance between pairwise vectors. Specifically, calculate the distance between the second word vector corresponding to each object topic and the first word vector corresponding to each target topic. For example, for the word vector corresponding to object topic n Calculate its word vector distance from each word vector in the word vector sequence corresponding to the target topic respectively. According to the word vector distance, determine whether there is a target topic in the target topic set that matches object topic n. Specifically, it can be implemented through the following formula:

[0154]

[0155]

[0156] Among them, threshold is a preset threshold; it can be seen from formula (5) that when the word vector distance is less than the preset threshold, it means that there is a target topic in the target topic set that matches object topic n. At this time, object topic n can be determined as the matching topic; otherwise, it means that there is no target topic in the target topic set that matches object topic n.

[0157] By generating word vectors corresponding to each topic, the target topics and object topics that may belong to different spaces can be mapped to the same space, so as to realize the description of each topic in a unified format, that is, the vector form representation of each topic, so as to facilitate the subsequent calculation of vector distance based on the same vector space. The method for generating word vectors in the present disclosure can be implemented by using existing related technologies for generating word vectors. For example, the Chinese word vector model can use ChineseWord2Vector, which will not be elaborated here.

[0158] In the above method, the matching topic corresponding to the candidate object can be determined according to the similarity between the object theme included in the candidate object and each target theme. Since the word vectors corresponding to each theme have been obtained through the word vector model before, the similarity between each theme can be determined by calculating the word vector distance between each theme. The calculation process of the word vector distance is fast and convenient, thus improving the efficiency of determining the matching topic. In addition, since the word vector can accurately describe the theme, the similarity calculated based on the word vector is more accurate, thereby improving the accuracy of determining the matching topic.

[0159] For the diversity index of each candidate object, it can be regarded as composed of at least one diversity index sub-item. For each candidate object, the number of diversity index sub-items is the same as the number of matching topics corresponding to the candidate object. It can be seen from this that the calculation of the diversity index sub-item needs to be based on the matching topic of the candidate object. Please refer to Figure 8 , which shows a method for calculating the diversity index sub-item. The method may include:

[0160] S810. Determine the number of candidate objects in the information retrieval result queue that include the matching topic.

[0161] S820. Calculate the diversity index sub-item corresponding to the matching topic based on the preset diversity index sub-item calculation formula and the number of candidate objects in the information retrieval result queue that include the matching topic.

[0162] According to the above content, for each candidate object, the corresponding matching topic can be determined. The matching topic is obtained by matching the object theme of the candidate object with the target theme. For the object theme n, when the target theme that matches the object theme n is found in the target theme set by calculating the word vector distance, although the two match, they may not be the same in the literal expression form. Therefore, a corresponding target theme identifier can be assigned to each target theme in the target theme set. For the object theme determined to be the matching topic, the target theme identifier of the target theme it matches can be marked for it. The target theme in the target theme set that matches the matching topic can be determined through the target theme identifier carried by the matching topic. Therefore, when determining the number of candidate objects in the information retrieval result queue that include the matching topic, the target theme identifier carried by the current matching topic can be determined first, and then the object theme carrying this target theme identifier can be searched in the information retrieval result queue, and the candidate object including the object theme carrying this target theme identifier can be found, so as to determine the number of candidate objects including the matching topic.

[0163] For any matching topic n, it corresponds to one diversity index sub-item. The calculation formula of the diversity index sub-item is as follows:

[0164]

[0165] Among them, 0 < α < 1, and α can generally be taken as 0.5. t is the index of the matching topic. For example, the t-th object topic of the candidate object is the matching topic; c t is the number of candidate objects containing the matching topic in the current information retrieval result queue, so that the diversity index sub-item corresponding to the matching topic n can be calculated.

[0166] S260. Determine the candidate objects whose diversity index meets the preset conditions as the retrieval result objects.

[0167] For the calculation of the diversity index of each candidate object, it can be carried out through the following formula:

[0168]

[0169] Among them, T is the number of object topics contained in any candidate object P, t is the index of the matching topic. For example, the t-th object topic of the candidate object is the matching topic; c t is the number of candidate objects containing the matching topic in the current information retrieval result queue; topic q is the set of target topics corresponding to the search term; topic p is the set of object topics contained in the candidate object. When the candidate object P corresponds to one matching topic, the diversity index sub-item corresponding to the matching topic can be directly used as the diversity index of the candidate object P; when the candidate object P corresponds to multiple matching topics, the diversity index sub-items corresponding to each matching topic can be summed to obtain the diversity index of the candidate object P.

[0170] In addition, after each theme matching, the number of candidate objects containing each matching theme in the information retrieval result queue can be updated accordingly, that is, the number of object topics matching each target theme in the information retrieval result queue is updated.

[0171] For the current matching topic, since there may already be candidate objects containing the current matching topic in the information retrieval result queue, that is, the current matching topic has already appeared in other previously selected candidate objects. When the matching topic appears again, its impact on the retrieval result will be affected by the number of times it has appeared, and the contribution of the matching topic to the diversity of the overall retrieval result is reduced. Therefore, the calculation of the diversity sub-item for the current matching topic needs to be based on the number of times it has appeared previously, further improving the accuracy of the calculation of the diversity index of each candidate object.

[0172] In this embodiment, the candidate object whose diversity index meets the preset condition is the candidate object with the largest diversity index, and the candidate object with the largest diversity index is the candidate object that contributes the most to the diversity of the overall retrieval result at present; when the number of candidate objects with the largest diversity index is one, directly determine this candidate object as the retrieval result object; when the number of candidate objects with the largest diversity index is greater than one, please refer to specifically Figure 9 , which shows a method for determining a retrieval result object, and the method may include:

[0173] S910. When the number of candidate objects with the largest diversity index is greater than one, determine the number of non-matching topics included in each candidate object with the largest diversity index.

[0174] S920. Determine the candidate object with the smallest number of non-matching topics as the retrieval result object.

[0175] Each candidate object may correspond to multiple object topics, and the multiple object indicators include matching topics and non-matching topics. When the number of candidate objects with the largest diversity index is greater than one, determine the candidate object with the smallest number of non-matching topics as the retrieval result object, so that the non-matching topics included in the information retrieval result presented to the user are as few as possible, which can also be regarded as having few invalid topics, thereby reducing the time for the user to browse invalid topics and improving the user experience.

[0176] S270. Add the retrieval result object to the tail of the information retrieval result queue, and delete the retrieval result object from the candidate object set.

[0177] At this time, the information retrieval result queue and the candidate object set can be updated respectively to facilitate entering the next round of loop.

[0178] S280. If the candidate object set is an empty set, output the information retrieval result queue.

[0179] Each time a loop is performed, a candidate object is taken out from the candidate object set and added to the tail of the information retrieval result queue until the candidate object set is an empty set; the finally generated information retrieval result queue is the queue with the largest retrieval result diversity, so that an information retrieval result with the largest diversity corresponding to the search term can be generated.

[0180] The method for processing information retrieval results proposed in the present disclosure can be specifically applied to the video search scenario in a short video community. At this time, the candidate objects are short videos; each short video can include corresponding attribute information, such as title text information. Further, the short video can be divided into several image frames by frame, and the text in each image frame can be recognized through OCR (Optical Character Recognition) technology, so that the text information contained in the short video can be obtained; the objects contained in each image frame can also be determined through image recognition technology; further, the voice information contained in the short video can be recognized through voice recognition technology, so that the voice information contained in the short video can be obtained. Based on the text information, image information, and voice information contained in the short video, the tag information of the short video and the content related to the relevant topics can be determined.

[0181] In the embodiments of the present disclosure, when specifically calculating the diversity index, it is implemented based on the diversity evaluation index α-NDCG. The implementation principle is that newly discovered topics are rewarded and already discovered topics are punished; the specific calculation process of α-NDCG can be illustrated by the following example.

[0182] The target topics included in the currently input search terms are {1, 2, 3, 4, 5}, and the sorting of the current information retrieval results is a, b, c, where:

[0183] Object a includes the object topics {1, 2},

[0184] Object b includes the object topics {3, 6, 7},

[0185] Object c includes the object topics {3, 4, 5},

[0186] For object a, it includes the target topics 1 and 2. Therefore, when calculating α-NDCG, it includes two items (1 - α)^0 and considering the sorting position is 1; for object b, it includes the target topic 3, and the target topic 3 has not appeared in the previous retrieval results. Therefore, when calculating α-NDCG, it includes one item (1 - α)^0 and considering the sorting position is 2; for object c, it includes the target topics 3, 4, 5. Since the target topic 3 has appeared once in the previous retrieval results and the target topics 4 and 5 have not appeared in the previous retrieval results, when calculating α-NDCG, it includes one item (1 - α) 1 and two items (1 - α)^0 and considering the sorting position is 3. Finally, the sum of the α-NDCG of objects a, b, and c is the diversity evaluation index α-NDCG of the final retrieval result.

[0187] The disclosed embodiments specifically adopt a topic matching method based on word embeddings and a candidate object ranking method based on greedy rules, and combine the calculation method of diversity indicators to process the candidate object set, thereby enhancing the diversity of information retrieval results.

[0188] The present disclosure obtains an information retrieval request including a search term, obtains a candidate object set based on the search term, and determines a target object set corresponding to the search term; based on the calculation result of the diversity index of each candidate object in the candidate object set, selects candidate objects whose diversity index meets preset conditions from the candidate object set and adds them to the end of the information retrieval result queue in sequence; when the candidate object set is an empty set, determines the information retrieval result queue currently containing multiple target objects as the information retrieval result corresponding to the search term and outputs it. The present disclosure can ensure that the top-ranked candidate objects in the returned information retrieval results contain as many different topics as possible, thereby improving the topic coverage of the retrieval results, thereby effectively covering the user's intent, reducing the redundancy of the retrieval results, enhancing the diversity of the retrieval results, and improving the accuracy of the information retrieval results.

[0189] The present disclosure also provides an information retrieval result processing device. Figure 10 , the device may include:

[0190] An information retrieval request obtaining unit 1010 is configured to obtain an information retrieval request, wherein the information retrieval request includes a search term;

[0191] The information retrieval request processing unit 1020 is configured to obtain a set of candidate objects according to the search term and determine a set of target topics corresponding to the search term;

[0192] The queue initialization unit 1030 is configured to initialize the information retrieval result queue;

[0193] The repetitive execution unit 1040 is configured to repeatedly call the following units until the candidate object set is empty:

[0194] The diversity index calculation unit 1050 is configured to calculate the diversity index of each candidate object in the candidate object set according to the object topics contained in each candidate object and the target topic set;

[0195] The search result object determining unit 1060 is configured to determine the candidate objects whose diversity indicators meet the preset conditions as search result objects;

[0196] An update unit 1070, configured to execute adding the retrieved result object to the tail of the information retrieval result queue and deleting the retrieved result object from the candidate object set;

[0197] An information retrieval result generation unit 1080, configured to execute outputting the information retrieval result queue if the candidate object set is an empty set.

[0198] In an exemplary embodiment, the diversity metric calculation unit 1050 includes:

[0199] A topic matching unit, configured to execute matching each object topic included in each candidate object with each target topic in the target topic set one by one to obtain a matching topic corresponding to each candidate object;

[0200] A diversity metric sub-item calculation unit, configured to execute calculating a diversity metric sub-item corresponding to the matching topic;

[0201] A diversity metric determination unit, configured to execute obtaining the diversity metric of each candidate object based on the diversity metric sub-item.

[0202] In an exemplary embodiment, the apparatus further includes:

[0203] A first word vector generation unit, configured to execute generating a first word vector corresponding to each target topic in the target topic set;

[0204] An object topic determination unit, configured to execute determining at least one object topic included in each candidate object in the candidate object set;

[0205] A second word vector generation unit, configured to execute generating a second word vector corresponding to each object topic.

[0206] In an exemplary embodiment, the topic matching unit includes:

[0207] A matching topic judgment unit, configured to execute judging whether each object topic included in each candidate object is a matching topic;

[0208] A matching result determination unit, configured to execute obtaining a matching topic corresponding to each candidate object based on the judgment result of each object topic of each candidate object.

[0209] In an exemplary embodiment, the matching topic judgment unit includes:

[0210] A word vector distance calculation unit, configured to calculate, one by one, the distance between the second word vector corresponding to each object topic and the first word vector corresponding to each target topic in the target topic set, to obtain a word vector distance set;

[0211] A comparison unit, configured to compare the smallest word vector distance in the word vector distance set with a preset threshold;

[0212] A matching topic determination unit, configured to determine each object topic as a matching topic when the smallest word vector distance is less than the preset threshold.

[0213] In an exemplary embodiment, the diversity metric sub-item calculation unit includes:

[0214] A first determination unit, configured to determine the number of candidate objects including the matching topic in the information retrieval result queue;

[0215] A first calculation unit, configured to calculate the diversity metric sub-item corresponding to the matching topic based on a preset diversity metric sub-item calculation formula and the number of candidate objects including the matching topic in the information retrieval result queue.

[0216] In an exemplary embodiment, the apparatus further includes:

[0217] A content recognition unit, configured to perform character recognition and image recognition on the content of each object in the information resource library, and determine the label information included in each object according to the recognition result;

[0218] The information retrieval request processing unit 1020 includes:

[0219] A search term tokenization unit, configured to tokenize the search term to obtain at least one search keyword;

[0220] A search object set determination unit, configured to respectively determine a search object set corresponding to each search keyword from the information resource library; wherein each search keyword is relevant to the label information of each search object in the corresponding search object set;

[0221] A candidate object set determination unit, configured to generate the candidate object set based on the search object set.

[0222] In an exemplary embodiment, the information retrieval request processing unit 1020 includes:

[0223] A search term analysis unit, configured to analyze the search term to determine a potential search intent corresponding to the search term;

[0224] A theme mining unit, configured to perform theme mining based on a potential search intention corresponding to the search term and generate a plurality of target themes corresponding to the search term;

[0225] A target theme set generation unit, configured to generate the target theme set based on the plurality of target themes corresponding to the search term.

[0226] In an exemplary embodiment, a candidate object whose diversity index meets a preset condition is a candidate object with the largest diversity index;

[0227] The retrieval result object determination unit 1060 includes:

[0228] A non-matching theme number determination unit, configured to determine the number of non-matching themes included in each candidate object with the largest diversity index when the number of candidate objects with the largest diversity index is greater than one;

[0229] A second determination unit, configured to determine the candidate object with the smallest number of non-matching themes as the retrieval result object.

[0230] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment related to the method, and will not be elaborated here.

[0231] In an embodiment of the present disclosure, a storage medium including instructions is further provided. When the instructions in the storage medium are executed by a processor of a server, the server is enabled to execute any of the methods in the above embodiments of the present disclosure. Optionally, the storage medium may be a non-temporary computer-readable storage medium. For example, the non-temporary computer-readable storage medium may be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0232] This embodiment further provides a server, and its structural diagram can be seen in Figure 11, the server 1100 can vary significantly due to differences in configuration or performance, and may include one or more central processing units (CPUs) 1122 (e.g., one or more processors) and a memory 1132, one or more storage media 1130 (e.g., one or more mass storage devices) for storing application programs 1142 or data 1144. Among them, the memory 1132 and the storage media 1130 can be transient storage or persistent storage. The programs stored in the storage media 1130 can include one or more modules (not shown in the figure), and each module can include a series of instruction operations on the device. Further, the central processing unit 1122 can be configured to communicate with the storage media 1130 and execute a series of instruction operations in the storage media 1130 on the device 1100. The device 1100 can also include one or more power supplies 1126, one or more wired or wireless network interfaces 1150, one or more input / output interfaces 1158, and / or one or more operating systems 1141, such as Windows Server TM , Mac OS X TM , Unix TM , Linux TM , FreeBSD TM and so on. Any of the above methods in this embodiment can be implemented based on the server shown in Figure 11 .

[0233] After considering the specification and practicing the invention disclosed herein, those skilled in the art will readily conceive of other embodiments of the present disclosure. The present disclosure is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include known common knowledge or conventional technical means in the technical field not disclosed by the present disclosure. The specification and embodiments are only to be considered as exemplary, and the true scope and spirit of the present disclosure are pointed out by the following claims.

[0234] It should be understood that the present disclosure is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present disclosure is only limited by the appended claims.

Claims

1. A method for processing information retrieval results, characterized in that Including: Obtain an information retrieval request, where the information retrieval request includes search terms; Obtain a candidate object set according to the search terms, and determine a target topic set corresponding to the search terms; The determining the target topic set corresponding to the search terms includes: analyzing the search terms to determine potential search intents corresponding to the search terms; performing topic mining based on the potential search intents corresponding to the search terms to generate multiple target topics corresponding to the search terms; generating the target topic set based on the multiple target topics corresponding to the search terms; Initialize an information retrieval result queue; Repeat the following steps until the candidate object set is an empty set: For each object topic included in each candidate object in the candidate object set, match each object topic with each target topic in the target topic set one by one to obtain a matching topic corresponding to each candidate object; calculate a diversity metric sub-item corresponding to the matching topic; obtain a diversity metric for each candidate object based on the diversity metric sub-item; Determine a candidate object whose diversity metric meets a preset condition as a retrieval result object; Add the retrieval result object to the end of the information retrieval result queue, and delete the retrieval result object from the candidate object set; If the candidate object set is an empty set, output the information retrieval result queue.

2. The information retrieval result processing method according to claim 1, wherein: After determining the target topic set corresponding to the search terms, it further includes: Generate a first word vector corresponding to each target topic in the target topic set; After obtaining the candidate object set according to the search terms, it further includes: Determine at least one object topic included in each candidate object in the candidate object set; Generate a second word vector corresponding to each object topic.

3. The information retrieval result processing method according to claim 2, wherein The step of, for each object topic included in each candidate object, matching each object topic with each target topic in the target topic set one by one to obtain a matching topic corresponding to each candidate object includes: Determine whether each object topic included in each candidate object is a matching topic; Obtain a matching topic corresponding to each candidate object based on the determination result of each object topic of each candidate object.

4. The information retrieval result processing method according to claim 3, characterized in that The determining whether each object topic included in each candidate object is a matching topic includes: Calculate the distance between the second word vector corresponding to each object topic and the first word vector corresponding to each target topic in the target topic set one by one to obtain a word vector distance set; Compare the smallest word vector distance in the word vector set with a preset threshold; When the smallest word vector distance is less than the preset threshold, determine each object topic as a matching topic.

5. The information retrieval result processing method according to claim 1, wherein The calculating the diversity metric sub-item corresponding to the matching topic includes: Determine the number of candidate objects including the matching topic in the information retrieval result queue; Calculate a diversity metric sub-item corresponding to the matching topic based on a preset calculation formula for the diversity metric sub-items and the number of candidate objects containing the matching topic in the information retrieval result queue.

6. The information retrieval result processing method according to claim 1, wherein Before obtaining the candidate object set according to the search term, the following steps are further included: Perform text recognition and image recognition on the content of each object in the information resource library, and determine the label information of each object according to the recognition results. Obtaining the candidate object set according to the search term includes: Segment the search term to obtain at least one search keyword. Respectively determine a search object set corresponding to each search keyword from the information resource library; wherein each search keyword has a correlation with the label information of each search object in the corresponding search object set. Generate the candidate object set based on the search object sets.

7. The information retrieval result processing method according to claim 1, characterized in that The candidate object whose diversity metric meets the preset condition is the candidate object with the largest diversity metric. Determining the candidate object whose diversity metric meets the preset condition as the retrieval result object includes: When the number of candidate objects with the largest diversity metric is greater than one, determine the number of non-matching topics included in each candidate object with the largest diversity metric. Determine the candidate object with the smallest number of non-matching topics as the retrieval result object.

8. An information retrieval result processing apparatus, characterized in that, Includes: An information retrieval request acquisition unit, configured to execute the acquisition of an information retrieval request, where the information retrieval request includes a search term. An information retrieval request processing unit, configured to execute obtaining a candidate object set according to the search term and determining a target topic set corresponding to the search term. The information retrieval request processing unit includes: a search term analysis unit, configured to execute the analysis of the search term to determine the potential search intent corresponding to the search term; a topic mining unit, configured to execute topic mining based on the potential search intent corresponding to the search term to generate multiple target topics corresponding to the search term; a target topic set generation unit, configured to execute generating the target topic set based on the multiple target topics corresponding to the search term. A queue initialization unit, configured to execute the initialization of the information retrieval result queue. A repeated execution unit, configured to execute repeated calls to the following units until the candidate object set is an empty set: A diversity metric calculation unit, configured to execute calculating the diversity metric of each candidate object according to the object topic included in each candidate object in the candidate object set and the target topic set; the diversity metric calculation unit includes: a topic matching unit, configured to execute, for each object topic included in each candidate object, matching each object topic with each target topic in the target topic set one by one to obtain a matching topic corresponding to each candidate object; a diversity metric sub-item calculation unit, configured to execute calculating the diversity metric sub-item corresponding to the matching topic; a diversity metric determination unit, configured to execute obtaining the diversity metric of each candidate object based on the diversity metric sub-item. A retrieval result object determination unit, configured to determine a candidate object whose diversity metric meets a preset condition as a retrieval result object; An update unit, configured to execute adding the retrieval result object to the tail of the information retrieval result queue and deleting the retrieval result object from the candidate object set; An information retrieval result generation unit, configured to execute outputting the information retrieval result queue if the candidate object set is an empty set.

9. The information retrieval result processing device according to claim 8, wherein The apparatus further includes: A first word vector generation unit, configured to generate a first word vector corresponding to each target theme in the target theme set; An object theme determination unit, configured to determine at least one object theme included in each candidate object in the candidate object set; A second word vector generation unit, configured to generate a second word vector corresponding to each object theme.

10. The information retrieval result processing apparatus according to claim 9, characterized in that, The theme matching unit includes: A matching theme judgment unit, configured to judge whether each object theme included in each candidate object is a matching theme; A matching result determination unit, configured to obtain a matching theme corresponding to each candidate object based on the judgment results of each object theme of each candidate object.

11. The information retrieval result processing apparatus according to claim 10, wherein The matching theme judgment unit includes: A word vector distance calculation unit, configured to calculate the distance between the second word vector corresponding to each object theme and the first word vector corresponding to each target theme in the target theme set one by one, to obtain a word vector distance set; A comparison unit, configured to compare the smallest word vector distance in the word vector distance set with a preset threshold; A matching theme determination unit, configured to determine each object theme as a matching theme when the smallest word vector distance is less than the preset threshold.

12. The information retrieval result processing apparatus according to claim 8, wherein The diversity metric sub-item calculation unit includes: A first determination unit, configured to determine the number of candidate objects including the matching theme in the information retrieval result queue; A first calculation unit, configured to calculate a diversity metric sub-item corresponding to the matching theme based on a preset diversity metric sub-item calculation formula and the number of candidate objects including the matching theme in the information retrieval result queue.

13. The information retrieval result processing device according to claim 8, characterized in that, The apparatus further includes: A content recognition unit, configured to perform character recognition and image recognition on the content of each object in the information resource library, and determine the tag information included in each object according to the recognition result; The information retrieval request processing unit includes: A search term word segmentation unit, configured to perform word segmentation on the search term to obtain at least one search keyword; A search object set determination unit, configured to respectively determine a search object set corresponding to each search keyword from the information resource library; where each search keyword is relevant to the tag information of each search object in the corresponding search object set; A candidate object set determination unit, configured to generate the candidate object set based on the search object set.

14. The information retrieval result processing apparatus according to claim 8, wherein The candidate object whose diversity metric meets the preset condition is the candidate object with the largest diversity metric; The retrieval result object determination unit includes: A non-matching topic quantity determination unit configured to, when the number of candidate objects with the maximum diversity index is greater than one, determine the number of non-matching topics included in each candidate object with the maximum diversity index; A second determination unit configured to determine the candidate object with the minimum number of non-matching topics as the retrieval result object.

15. A server, characterized in that, Comprising: A processor; A memory for storing instructions executable by the processor; Wherein, the processor is configured to execute the instructions to implement the information retrieval result processing method according to any one of claims 1 to 7.

16. A storage medium, characterized in that, When the instructions in the storage medium are executed by the processor of the server, the server is enabled to execute the information retrieval result processing method according to any one of claims 1 to 7.

Images