Search term package determination method and apparatus, electronic device, and storage medium

Abstract

The application discloses a search word bag determination method and device, electronic equipment and a storage medium. The method comprises the following steps: obtaining a plurality of search words and a plurality of recall results corresponding to the plurality of search words; determining a search index of each search word based on the plurality of recall results, wherein the search index is a value index related to the search between each search word and a first search word, and the first search word is a search word remaining after removing each search word from the plurality of search words; determining an industry word used for representing a target industry in the plurality of search words; and determining a search word bag of the target industry based on the search index of the industry word and a second search word, wherein the second search word is a search word remaining after removing the industry word from the plurality of search words. The method can automatically generate an industry search word bag and reduce the generation cost of the industry word bag.

Description

Technical Field

[0001] This invention relates to the field of search term package determination technology, and specifically to a search term package determination method, apparatus, electronic device, and storage medium. Background Technology

[0002] To facilitate retrieval, each industry typically defines a set of search terms. When users search using these terms, data corresponding to that industry is displayed for them to browse. Currently, a common method for defining these terms is to manually define the keyword set corresponding to each industry and then perform fuzzy matching. For example, for the men's face cream industry, the manual definition method usually uses the words "men" and "face cream" as conditions to perform fuzzy matching on search terms.

[0003] However, this method requires extensive preliminary work to acquire and analyze industry-related data in order to improve the accuracy of manually defined search terms. This work demands a significant investment of manpower, resulting in high labor costs and relatively low efficiency in acquiring industry search term packages. Summary of the Invention

[0004] To address the aforementioned problems in the prior art, this application provides a method, apparatus, electronic device, and storage medium for determining search term packages, which can automatically generate industry search term packages, reduce the generation cost of industry term packages, and improve generation efficiency.

[0005] In a first aspect, embodiments of this application provide a method for determining search term packages, the method comprising:

[0006] Retrieve multiple search terms and multiple recall results corresponding to the multiple search terms;

[0007] Based on multiple recall results, the search metrics for each search term are determined. The search metrics are the value indicators of the search relevance between each search term and the first search term, which is the search term remaining after removing each of the multiple search terms.

[0008] Identify industry terms used to characterize the target industry from multiple search terms;

[0009] Based on the search metrics of industry terms and secondary search terms, a search term package for the target industry is determined. The secondary search term is the search term remaining after removing the industry terms from a pool of search terms.

[0010] In one possible implementation, the search metrics for each search term include the correlation coefficient between each search term and the remaining first search terms, as well as the search popularity of each search term.

[0011] Based on multiple recall results, the search metrics for each search term were determined, including:

[0012] Based on the access records of multiple recall results, determine the correlation coefficient between any two search terms;

[0013] The search popularity of each search term is determined based on the correlation coefficient between each search term and the first search term.

[0014] In one possible implementation, based on access records from multiple recall results, the relevance coefficient between any two search terms is determined, including:

[0015] Based on the access records of multiple recall results, determine the total number of accesses for each recall result and the number of associated accesses between each recall result and the search term corresponding to that recall result. The associated access count is the number of times the recall result is accessed when the recall result is searched using the corresponding search term.

[0016] The correlation coefficient between any two search terms is determined based on the total number of visits for each recall result and the number of associated visits between each recall result and the search term corresponding to that recall result.

[0017] In one possible implementation, a search term package for the target industry is determined based on search metrics of industry terms and second search terms, including:

[0018] Based on the correlation coefficient between industry terms and second search terms, and the search popularity of industry terms and second search terms, multiple industry search terms are identified in the second search terms. Among them, the correlation coefficient between each industry search term and the industry term is greater than the first threshold, and the ratio of the search popularity of the industry term to the search popularity of each industry search term is greater than the second threshold.

[0019] The industry term and multiple industry search terms are used as a search term package for the corresponding industry.

[0020] In one possible implementation, industry terms for characterizing the target industry are identified from a plurality of search terms, including:

[0021] Multiple search terms are matched with a preset industry dictionary to determine industry terms, which contains a number of pre-recorded industry terms.

[0022] In one possible implementation, before matching multiple search terms with a preset industry dictionary to determine industry terms, the method further includes:

[0023] Retrieve multiple indivisible minimum business categories from the business category table;

[0024] Use the product name of each minimum business category as the industry term for the corresponding industry of each minimum business category;

[0025] Generate an industry dictionary based on multiple industry terms.

[0026] In one possible implementation, multiple recall results are data that has been searched and effectively accessed through any one of multiple search terms, wherein effective access refers to the access time for each recall result being greater than the effective access time threshold corresponding to the data type of the recall result.

[0027] In one possible implementation, the method further includes:

[0028] Obtain historical search term packages for the target industry;

[0029] Based on the search term package and the historical search term package, multiple first-differentiating words, multiple second-differentiating words, and multiple common words are identified. Each first-differentiating word is a word that exists in the search term package but not in the historical term package, and each second-differentiating word is a word that exists in the historical search term package but not in the search term package.

[0030] By filtering the first and second difference words, multiple target words are obtained;

[0031] Use multiple target terms and multiple common terms as the industry-specific search term package.

[0032] In one possible implementation, the first and second difference words are filtered to obtain multiple target words, including:

[0033] Based on the similarity between each first difference word and each second difference word, multiple difference word groups are determined among multiple first difference words and multiple second difference words, wherein each difference word group includes a third difference word and a fourth difference word, multiple first difference words include a third difference word, multiple second difference words include a fourth difference word, and the similarity between the third difference word and the fourth difference word is greater than a third threshold.

[0034] Each difference phrase is filtered to obtain multiple fifth difference words;

[0035] The remaining difference words after removing the difference words from multiple difference word groups from multiple fifth difference words, multiple first difference words, and multiple second difference words are taken as multiple target words.

[0036] In one possible implementation, each difference word group is filtered to obtain multiple fifth difference words, including:

[0037] Determine the usage popularity of the third and fourth differentiating words in each differentiating phrase within a preset time period;

[0038] If the usage popularity of the fourth difference word is greater than or equal to that of the third difference word, the third and fourth difference words will be identified as the retained words for each difference word group.

[0039] If the usage popularity of the fourth difference word is less than that of the third difference word, determine the difference between the usage popularity of the third difference word and the usage popularity of the fourth difference word;

[0040] If the difference between the usage popularity of the third differentiating word and the usage popularity of the fourth differentiating word is greater than the fourth threshold, the third differentiating word will be determined as a retained word for each differentiating word group.

[0041] If the difference between the usage popularity of the third difference word and the usage popularity of the fourth difference word is less than or equal to the fourth threshold, the third difference word and the fourth difference word are determined as the retained words for each difference word group;

[0042] The retained words corresponding to multiple different word groups are treated as multiple fifth different words.

[0043] In one possible implementation, the method further includes:

[0044] Retrieve multiple brand terms from the search term package;

[0045] Based on the search popularity of each brand term, multiple brand terms are sorted to obtain the brand ranking of the industry corresponding to the search term package.

[0046] In one possible implementation, the method further includes:

[0047] Get the search count for each search term in the search term package;

[0048] Based on the number of searches for each search term, determine the search trends for the corresponding industry for the search term package.

[0049] Secondly, embodiments of this application provide a search term bag determination device, comprising:

[0050] The acquisition module is used to acquire multiple search terms and multiple recall results corresponding to the multiple search terms;

[0051] The analysis module is used to determine the search metrics for each search term based on multiple recall results. The search metrics are the value indicators of the search relevance between each search term and the first search term, which is the search term remaining after removing each of the multiple search terms.

[0052] The determination module is used to identify industry terms that characterize the target industry from multiple search terms;

[0053] The processing module is used to determine the search term package for the target industry based on the search metrics of industry terms and second search terms. The second search term is the search term remaining after removing industry terms from a plurality of search terms.

[0054] Thirdly, embodiments of this application provide an electronic device, including: a processor connected to a memory for storing a computer program, and the processor for executing the computer program stored in the memory to cause the electronic device to perform the method as described in the first aspect.

[0055] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that causes a computer to perform the method as described in the first aspect.

[0056] Fifthly, embodiments of this application provide a computer program product, the computer program product including a non-transitory computer-readable storage medium storing a computer program, and a computer operable to perform the method as described in the first aspect.

[0057] Implementing the embodiments of this application has the following beneficial effects:

[0058] In this embodiment, multiple search terms and corresponding recall results are obtained. Based on these recall results, a search metric is determined that identifies the search relevance value between each search term and the remaining first search term among the multiple search terms. Then, industry terms representing the target industry are identified from the multiple search terms. Based on the search metric of the industry terms and the remaining second search terms among the multiple search terms, a search term package for the target industry is determined. Specifically, by analyzing multiple recall results of multiple search terms containing industry terms in practical applications, a search metric is quantified that identifies the true search value of each search term in practical applications compared to other search terms. Based on this, for an industry term among multiple search terms, its true search value compared to other search terms can be determined using this search metric, thereby identifying the search term package for the corresponding industry among the multiple search terms. Thus, the automatic generation of industry search term packages can be achieved without manual intervention, greatly reducing the generation cost and improving the efficiency of search term package generation. At the same time, by using real search terms and recall results, the generated search term package is more closely related to the user's real usage scenarios, giving the generated search term package higher search value and accuracy. Attached Figure Description

[0059] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0060] Figure 1 A schematic diagram of a search term package determination system provided for embodiments of this application;

[0061] Figure 2 A schematic diagram illustrating a method for determining search term packages in a sharing platform scenario, provided as an embodiment of this application;

[0062] Figure 3 A flowchart illustrating a method for determining a search term package as provided in this application.

[0063] Figure 4 A schematic diagram of a business category provided for the implementation of this application;

[0064] Figure 5 A schematic diagram illustrating a method for determining a more precise search term package for a target industry based on a search term package and a historical search term package, provided for implementation of this application;

[0065] Figure 6 A schematic diagram illustrating the filtering of different words in a phrase as provided in this application.

[0066] Figure 7 A flowchart illustrating another method for determining search term packages provided in this application;

[0067] Figure 8 A schematic diagram illustrating how to view industry brand rankings for the purposes of this application;

[0068] Figure 9 A flowchart illustrating another method for determining search term packages provided in this application;

[0069] Figure 10 A schematic diagram illustrating search trends provided for an embodiment of this application;

[0070] Figure 11 A functional module block diagram of a search term package determination device provided for embodiments of this application;

[0071] Figure 12 This is a schematic diagram of the structure of an electronic device provided for an embodiment of this application. Detailed Implementation

[0072] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0073] The terms "first," "second," "third," and "fourth," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0074] In this document, the term "implementation" means that a specific feature, result, or characteristic described in connection with an implementation may be included in at least one implementation of this application. The appearance of this phrase in various places in the specification does not necessarily refer to the same implementation, nor is it a separate or alternative implementation mutually exclusive with other implementations. It will be explicitly and implicitly understood by those skilled in the art that the implementations described herein can be combined with other implementations.

[0075] First, the relevant terms used in this application will be explained:

[0076] Search terms: These are the words that users enter into the search box to perform a search.

[0077] Industry terms: These are predefined terms used to characterize an industry.

[0078] Correlation coefficient: refers to the degree of correlation between two search terms in terms of search power and value.

[0079] Search popularity: refers to the frequency with which a search term is used among multiple search terms retrieved at once.

[0080] Secondly, see Figure 1 , Figure 1 This is a schematic diagram of a search term package determination system provided for an embodiment of this application.

[0081] For example, the search term determination system may include a search term determination device and a database. The search term determination device may be a server, for example, a standalone server, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. This application does not specifically limit its scope. The database can also be a server, or a storage device that provides data storage services, such as: read-only memory (ROM) or other types of static storage devices that can store static information and instructions, random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, and this application does not specifically limit this as well.

[0082] Specifically, the search term packet determination device periodically retrieves multiple search terms received by the platform in the current period from the database, along with the corresponding recall results. For example, a one-day period can be used, with the database recording data from all search tasks received by the platform within a day. The search term packet determination device retrieves new data from the database daily, analyzes this data, obtains the search terms used in each search task and the recall results retrieved for that task, and associates the search terms and recall results under the same task. After analyzing all search tasks, the recall results corresponding to the same search terms extracted from each search task are merged, ensuring that no two search terms in the final result are identical, thus obtaining multiple search terms for the current period and their corresponding recall results.

[0083] Then, the search term package determination device determines the search metric for each search term based on multiple recall results. Specifically, the search metric is used to identify the search relevance of each search term to the remaining first search term in the current search cycle. The remaining first search term can be understood as the search term remaining after removing each of the multiple search terms. Through this search metric, the true search value relationship between the multiple search terms collected in the current cycle can be quantified.

[0084] Finally, the search term package determination device identifies industry terms representing the target industry from among multiple search terms. Based on the search metrics of these industry terms and the remaining second search terms among the multiple search terms, it determines several words that are relevant to the industry terms in search and have certain value, thus obtaining a search term package for the target industry. Similarly, the remaining second search terms mentioned here can be understood as the search terms remaining after removing the industry terms from among the multiple search terms.

[0085] Finally, it should be noted that the search term package determination method provided in this application can be applied to various search platforms, databases, sharing platforms, and other scenarios where data can be obtained through search.

[0086] Taking a sharing platform scenario as an example, such as Figure 2 As shown, users can search for relevant content on the platform by entering their search query in the search box on the platform's search page. After receiving the search query, the platform analyzes it, compares the search terms with various content items on the platform, and displays content related to those search terms to the user.

[0087] Specifically, the search statement can be one or more words or a short phrase containing multiple words. When the search statement is one or more words, these words can be directly used as the search terms for this search. When the search statement is a short phrase containing multiple words, the multiple words in the short phrase can be extracted through word segmentation and then used as the search terms for this search.

[0088] In this implementation, the search content displayed to the user can be used as the recall results for this search. The platform will record these recall results and the user's access to each recall result. After the user finishes the search, the search terms, recall results, and access records of the recall results are associated and stored in the database as data for a search task.

[0089] This demonstrates that search terms play a crucial role in whether users can successfully find the content they want. The more accurately a search term retrieves the content a user needs, the higher its search capability value within the corresponding industry. Therefore, by accurately quantifying the value of each search term, multiple high-value search terms within a given industry can be obtained as a search term package for that industry. Based on this, this application provides a search term package determination method, apparatus, electronic device, and storage medium that can quantify the search capability value of the aforementioned search terms and automatically generate an industry-specific search term package.

[0090] Specifically, the search term package determination device can periodically retrieve search task data from the sharing platform in the previous period from the database, and then extract this data to obtain multiple search terms received by the sharing platform in the previous period and multiple recall results corresponding to these search terms. Then, based on the multiple recall results, a search metric is determined to characterize the search relevance value between each search term and the remaining first search terms, and this search metric is used as a quantitative indicator of the search capability value of each search term. Finally, industry terms representing the industry are identified from the multiple search terms, and based on the search metric between these industry terms and the remaining second search terms, the search term package for that industry is determined.

[0091] The following will use the scenario of a sharing platform as an example to illustrate the method for determining search term packages proposed in this application. The methods for determining search term packages in other scenarios are similar to those in the scenario of a sharing platform, and will not be described in detail here.

[0092] See Figure 3 , Figure 3 This is a flowchart illustrating a method for determining search term packages according to an embodiment of this application. The method is applied to the search term package determining device in the above embodiment and may include the following steps:

[0093] 301: Retrieves multiple search terms and multiple recall results corresponding to the multiple search terms.

[0094] In this embodiment, the search term packet determination device periodically retrieves multiple search terms received by the platform in the current period from the database according to a preset cycle, as well as multiple recall results corresponding to the multiple search terms. For example, a one-day cycle can be used, and the search term packet determination device retrieves the search data of the previous day from the database every other day, analyzes it, and obtains multiple search terms and multiple recall results corresponding to the multiple search terms.

[0095] Specifically, when a user performs a search task on the platform, the platform records the search query entered by the user, the multiple retrieved results obtained through that search query, and the user's access history for each retrieved result. The search query can be one or more words, or a short phrase containing multiple words. After receiving the search query, the platform can extract the words to obtain one or more search terms corresponding to the search task. In short, when the search query is a single word, that word can be directly used as the search term for this search task. For example, if the user enters the search query "facial cleanser," then the search term for this search task is "facial cleanser." When the search query consists of multiple words, each word can be separated by spaces. For example, if the user enters the search query "men's oil-control facial cleanser," then this search task has three search terms: "men," "oil-control," and "facial cleanser." When the search query is a short phrase containing multiple words, the phrase can be segmented to extract words with practical meaning as search terms. For example, if the user enters the search query "refreshing and oil-removing men's facial cleanser", after segmentation, there are 4 search terms for this search task: "refreshing", "oil-removing", "men's", and "facial cleanser".

[0096] In this implementation, all data retrieved using a search query in a single search task can be considered as the recall results for one or more search terms corresponding to that search query. For example, if the search query "men's oil-control facial cleanser" yields 10 results, these 10 results are the 10 recall results corresponding to the search terms "men," "oil-control," and "facial cleanser." Simultaneously, the platform records the user's access to each recall result as an access record for that result. This access record includes the search terms used, the number of accesses, and the duration of the access. For example, if a user clicks on only recall result 5 out of the 10 recall results and browses it for 10 seconds, the access record for recall result 5 would be: one access using the search terms "men," "oil-control," and "facial cleanser," lasting 10 seconds. The remaining 9 recall results, since they were not accessed, have empty access records.

[0097] In an optional implementation, only data with valid access can be recorded as the recall result to reduce the demand on computing power. Specifically, different valid access time thresholds can be set for different types of data. When the access time for a piece of data exceeds the access time threshold for the corresponding type, the data can be considered validly accessed. For example, for text and image data, the valid pre-reading time threshold can be set to 3 seconds, and for video data, the valid reading time threshold can be set to 5 seconds. Based on this, using the example of the search term "men's oil-removing facial cleanser" and 10 search data, if a user accesses data 1, data 3, data 5, and data 10 of the 10 search data respectively, where data 1 is text and image data viewed for 10 seconds, data 3 is video data viewed for 2 seconds, data 5 is video data viewed for 20 seconds, and data 10 is text and image data viewed for 1 second, then only data 1 and data 5 have access times exceeding the valid access time threshold corresponding to their data types. Therefore, among the 10 search data, data 1 and data 5 are the validly accessed data. For this, the search terms "men's," "oil removal," and "facial cleanser" yielded two recall results, data 1 and data 5, respectively. Therefore, only data with valid access needs to be recorded, significantly reducing the amount of data to be recorded, lowering the computational power requirement, and improving recording efficiency. Similarly, for each recall result, its access history should also be recorded and synchronously stored in the database. To better illustrate the advantages of the search term package determination method provided in this application, the recall results mentioned in the following embodiments will be assumed to be validly accessed data unless otherwise specified.

[0098] Therefore, in this embodiment, the daily search data on the platform is recorded in the database in the form of search tasks. After the search term package determination device obtains the search data for one day, it associates the search terms and recall results recorded in each search task with the access records of each recall result for statistical analysis. Specifically, firstly, the search terms in each search task are counted, and the recall results corresponding to each search term are determined. Then, the recall results corresponding to the same search terms among the counted search terms are merged. That is, if there are identical recall results among the recall results corresponding to the same search terms, the access counts of those recall results are accumulated and merged into one recall result.

[0099] For example, for search task 1, the corresponding search terms are search term 1 and search term 2; for search task 2, the corresponding search term is search term 1. Since two identical search terms 1 appear, the recall results corresponding to these two search terms 1 need to be merged. Specifically, the recall results for search term 1 in search task 1 are: Recall result 1 accessed 2 times, recall result 2 accessed 1 time, and recall result 3 accessed 1 time; the recall results for search term 1 in search task 2 are: Recall result 3 accessed 2 times and recall result 4 accessed 1 time. After merging, the final recall results for search term 1 are: Recall result 1 accessed 2 times, recall result 2 accessed 1 time, recall result 3 accessed 3 times, and recall result 4 accessed 1 time.

[0100] Based on this, by extracting and merging the recall results corresponding to the same search terms for each search task, we can obtain multiple search terms received by the platform within the period and multiple recall results corresponding to those multiple search terms.

[0101] 302: Based on multiple recall results, determine the search metrics for each search term.

[0102] In this embodiment, the search metric is a value indicator related to the search between each search term and the remaining first search term among multiple search terms. For example, the search metric for each search term may include the correlation coefficient between the search term and the first search term, and the search popularity of the search term. The correlation coefficient characterizes the degree of correlation between two search terms in terms of search capability and value, and the search popularity characterizes the frequency of use of the search term among multiple search terms acquired in the same period. The methods for determining the correlation coefficient and search popularity will be explained below.

[0103] In this embodiment, the relevance coefficient between any two search terms can be determined based on the access records of multiple recall results. Specifically, by using the access records of multiple recall results, the total number of visits to each recall result and the number of associated visits between each recall result and the search term corresponding to that recall result can be determined. The associated visit count refers to the number of times the recall result is accessed when it is retrieved using the corresponding search term.

[0104] As described in step 301, after merging, multiple distinct search terms within a given period and multiple recall results corresponding to each search term can be obtained. This correspondence is determined through the search task; that is, a correspondence only exists if the search term and the recall result originate from the same search task record. Furthermore, the fact that the search term and the recall result originate from the same search task indicates that the recall result was retrieved using that search term. Therefore, by statistically analyzing the access records of the recall results corresponding to each search term, the associated access count between each recall result and its corresponding search term can be determined. Then, by summing the associated access counts of each recall result under different search terms, the total access count of the recall result can be obtained.

[0105] For example, continuing with the example of search term 1 above, the corresponding recall results and the access records for each recall result are as follows: Recall result 1 was accessed 2 times, recall result 2 was accessed 1 time, recall result 3 was accessed 3 times, and recall result 4 was accessed 1 time. Therefore, for search term 1, the number of times it is associated with recall result 1 is 2, which can be denoted as C. (搜索词1，召回结果1) =2; its associated access count with recall result 2 is 1, which can be denoted as C. (搜索词1，召回结果2) =1; its associated access count with recall result 3 is 3, which can be denoted as C. (搜索词1，召回结果3) =3; its associated access count with recall result 4 is 1, which can be denoted as C. (搜索词1，召回结果4) =1.

[0106] Meanwhile, for recall result 1, there are three corresponding search terms: search term 1, search term 2, and search term 3. The number of visits associated with each of these three search terms is as follows: C (搜索词1，召回结果1) =2, C (搜索词2，召回结果1) =1 and C (搜索词3，召回结果1) =3. Therefore, the total number of visits for recall result 1 is the sum of the visits for these three related results: 6 times, which is C. 召回结果1 =6.

[0107] In this embodiment, after determining the total number of visits for each recall result and the associated visit count between each recall result and its corresponding search term, the correlation coefficient between any two search terms can be determined based on these figures. Specifically, this correlation coefficient can be expressed using formula ①:

[0108]

[0109] Where kwi represents the i-th search term among multiple search terms, kwj represents the j-th search term among multiple search terms, σ(kwi, kwj) represents the correlation coefficient between the i-th and j-th search terms, and C (kwi，k) C represents the number of associated visits between the i-th search term and the recall result k. (kwj，k) The number of associated visits between the j-th search term and the recall result k, C k The total number of visits to the recall result k, where n represents the number of recall results.

[0110] It can be seen that the more identical and effectively accessed recall results two search terms can retrieve, the higher the similarity between the two search terms in terms of search performance, and the more similar their values ​​are in terms of value. Quantitatively, this is expressed as a larger product of the number of associated visits corresponding to both terms under the same recall result. Therefore, by summing the products of the number of associated visits across all recall results, a larger value indicates a higher similarity between the two search terms in terms of search performance, and a more similar value, thus indicating a higher correlation between the two search terms in terms of search capability and value. To prevent the calculated result from being too large, before summing, the product of the number of associated visits under the same recall result is divided by the total number of visits for that recall result; after summing, the sum is divided by the sum of the total number of visits for all recall results. This constrains the value of the quantified result, preventing it from becoming excessively large, thereby simplifying subsequent calculations without affecting the meaning of the quantified value.

[0111] In this embodiment, after determining the correlation coefficient between any two search terms, the search popularity of each search term can be determined based on the correlation coefficient between each search term and the remaining first search terms. Specifically, the search popularity of each search term can be expressed by formula ②:

[0112]

[0113] Among them, I kwi represents the search popularity of the i-th search term, and m represents the number of search terms.

[0114] It can be seen that the more non-zero correlation coefficients a search term has with other search terms in the same period, the more overlap there is between the retrieved results of that search term and the retrieved results of other search terms. Therefore, the larger the sum of the correlation coefficients between the search term and other search terms, the greater the overlap between the retrieved results of that search term and those of other search terms, the stronger the likelihood that the search term will substitute for other search terms, and the higher its search value and search popularity within the current period.

[0115] In an optional implementation, the sum of the correlation coefficients between the search term and all other search terms in the same period that have a non-zero correlation coefficient can also be calculated to simplify the calculation.

[0116] 303: Identify industry terms used to characterize the target industry from among multiple search terms.

[0117] In this implementation, "industry" broadly refers to a business category. A business category can be further subdivided into multiple sub-industries from its overall industry classification, ultimately differentiating into the smallest indivisible business category, called a "leaf industry." For example... Figure 4 As shown, for the beauty and cosmetics business category, the overall industry is "beauty and cosmetics," which can be further subdivided into many sub-industries such as: "makeup," "hair care," "personal care," "beauty devices," "beauty tools," "beauty foods," "skincare," and "beauty services." Within the sub-industry "makeup," the next level of sub-industries can include "foundation makeup," "eye makeup," "lip makeup," and "blush." ​​The sub-industry "eye makeup" further subdivides into inseparable leaf-like sub-industries, such as "eyeshadow," "eyeliner," and "mascara."

[0118] In this implementation, "industry" mainly refers to these indivisible leaf industries, as they belong to the smallest, indivisible business categories. Their corresponding industry terms are mostly singular, namely the product names of the industry itself. Therefore, when calculating search term packages for these leaf industries, since they only have one corresponding industry term, the calculation is more efficient and has less error in actual calculations. For the parent industry of non-leaf industries within a business category, the calculation results of its subordinate sub-industries can be summed to obtain the calculation result for that parent industry.

[0119] Based on this, in this embodiment, after determining the search popularity of each search term and the correlation coefficient between any two search terms, the industry-specific terms corresponding to the industry can first be identified from multiple search terms. For example, multiple search terms can be matched with a preset industry dictionary to determine whether any of the search terms contain industry-specific terms. Specifically, the industry dictionary pre-records multiple industry-specific terms; if any of the search terms contain a term recorded in the industry dictionary, then that term can be identified as an industry-specific term.

[0120] In this embodiment, the industry dictionary can be predetermined. Specifically, multiple indivisible minimum business categories can be obtained from the above-mentioned business category table. The product name of each minimum business category is used as the industry term for the corresponding industry of each minimum business category, and an industry dictionary is generated based on multiple industry terms.

[0121] 304: Based on search metrics of industry terms and secondary search terms, determine the search term package for the target industry.

[0122] In this embodiment, after identifying the industry terms, multiple industry search terms can be determined from the remaining second search terms based on the correlation coefficient between the industry terms and the remaining second search terms among the multiple search terms, as well as the search popularity between the target search term and the second search terms. Specifically, the correlation coefficient between each industry search term and the other industry terms is greater than a first threshold, and the ratio of the search popularity of the industry term to the search popularity of each industry search term is greater than a second threshold. In other words, each industry search term satisfies the requirements of the following formula ③:

[0123]

[0124] Where kw represents an industry term, SET represents the search term package for the target industry, kwg represents any single industry search term, σ(kw, kwg) represents the correlation coefficient between the industry term and any single industry search term, and I kw Indicating the search popularity of industry-specific terms, I kwg This represents the search popularity of any industry search term, where s represents the first threshold and r represents the second threshold.

[0125] In this implementation, the first threshold s can typically be set to 0, meaning that search terms with a correlation coefficient greater than 0 with industry terms can be considered to have passed the first step of screening. Specifically, this condition is used to determine which search terms in the same period have a certain correlation with industry terms in terms of search.

[0126] Meanwhile, for industry-specific search term packages, users typically include industry-specific terms when searching for content within that industry to anchor their search scope and avoid finding overly niche results. Therefore, industry-specific terms should have the highest search popularity within the search term package, with a significant gap compared to other search terms. Based on this, in this implementation, for search terms that pass the initial screening in the first step, it is necessary to calculate the ratio of the search popularity of the industry-specific terms to the search popularity of the search terms that passed the initial screening in the first step. Searches with a ratio greater than a second threshold are selected for the second, more refined screening.

[0127] Therefore, some users may accidentally click on data unrelated to their original search intent, resulting in erroneous recall results. When calculating the relevance coefficient, a non-zero correlation coefficient might appear between two search terms that shouldn't be associated, allowing the term to pass the initial screening. Since this term doesn't actually belong to the industry's terminology package, the difference in search popularity between this term and other industry terms might not be significant. Based on this, by calculating the ratio of search popularity between industry terms and candidate search terms, these erroneous terms can be filtered out, leading to a more accurate terminology package.

[0128] In this embodiment, the specific value of the second threshold can be determined by factors such as the number of search terms that the search term package needs to be included and the value of the final generated search term package.

[0129] In this embodiment, after identifying multiple industry search terms, these multiple industry search terms and industry terms can be used as a search term package for the industry to which the target search term belongs. That is, the search term package for this industry consists of the industry term itself and multiple industry search terms identified through that industry term.

[0130] In the optional real-time mode, after obtaining the search term package for the target industry, it can also be determined whether a historical search term package exists for that industry. This historical search term package can be determined by the search term package determination device in previous cycles. If a historical search term package exists, it can be acquired, and a more precise search term package for the target industry can be determined based on the search term package and the historical search term package.

[0131] Specifically, such as Figure 5 As shown, firstly, based on the search term package and historical search term packages, multiple first-differentiating words, multiple second-differentiating words, and multiple common words can be identified. Each first-differentiating word is a word present in the search term package but not in the historical term package, and each second-differentiating word is a word present in the historical search term package but not in the current search term package. Then, the first-differentiating words and second-differentiating words can be filtered to obtain multiple target words. Finally, the multiple target words and multiple common words are used as the search term package for the corresponding industry.

[0132] In this embodiment, the similarity between each first differing word and each second differing word can be calculated, and multiple differing word groups can be determined from the multiple first differing words and multiple second differing words. Specifically, each differing word group includes two search terms from the search term package and the historical search term package, respectively: a third differing word and a fourth differing word. Multiple first differing words include the third differing word, multiple second differing words include the fourth differing word, and the similarity between the third differing word and the fourth differing word is greater than a third threshold. In short, two differing words with a similarity greater than the third threshold from the first and second differing words are grouped into a differing word group. For two differing words in a differing word group, since their similarity is high, in order to ensure the conciseness and value of the final search term package, it is necessary to filter these highly similar differing words and retain the words with higher value.

[0133] Specifically, such as Figure 6As shown, firstly, the usage popularity of the third and fourth differentiating words in each differentiating word group within a preset time period can be determined. This usage popularity can be the frequency with which the differentiating word is used in searches within the preset time period. For example, the number of times the third and fourth differentiating words were used in searches in the past month, as well as the total number of searches for all search terms in the past month, can be counted. Then, the ratio of the number of times the third differentiating word was used in searches to the total number of searches can be used as the usage popularity of the third differentiating word. Similarly, the ratio of the number of times the fourth differentiating word was used in searches to the total number of searches can be used as the usage popularity of the fourth differentiating word.

[0134] Then, if the usage popularity of the fourth differentiating term is greater than or equal to that of the third differentiating term, it indicates that the fourth differentiating term in the historical search term package still has a high usage frequency within the preset time period. Meanwhile, the third differentiating term, as an industry search term identified in this period, also has high timeliness. Therefore, both the third and fourth search terms have high search value for this industry, and thus, the third and fourth differentiating terms can be identified as retained terms for each differentiating term group.

[0135] If the usage popularity of the fourth differentiating term is less than that of the third differentiating term, it indicates that the fourth differentiating term in the historical search term package was used less frequently than the third differentiating term identified in the current period. The third differentiating term may be a newly emerging word used to replace the fourth differentiating term, thus leading to a decrease in the usage popularity of the fourth differentiating term. To address this, the difference between the usage popularity of the third and fourth differentiating terms can be further determined, and the specific situation can be determined based on the magnitude of this difference.

[0136] Specifically, if the difference in usage popularity between the third and fourth differentiating words is greater than the fourth threshold, it indicates that the usage frequency of the third differentiating word is much higher than that of the fourth differentiating word. Therefore, the third differentiating word is more likely to be an emerging word used to replace the fourth differentiating word. Since the third and fourth differentiating words are highly similar, only the third differentiating word with higher usage popularity can be selected as the retained word for each differentiating word group, while the fourth differentiating word with significantly lower usage popularity can be filtered out. If the difference in usage popularity between the third and fourth differentiating words is less than or equal to the fourth threshold, it indicates that the difference in usage frequency between the two is not significant. The third differentiating word is less likely to be an emerging word used to replace the fourth differentiating word, and both the third and fourth differentiating words can be selected as the retained words for each differentiating word group.

[0137] Finally, the retained words corresponding to multiple differential word groups are taken as multiple fifth differential words, and together with the differential words remaining after removing the differential words from multiple differential word groups from multiple first differential words and multiple second differential words, they are taken as multiple target words. Thus, the search term package determined in the current period is merged with the historical search term package, and industry search terms with similar alternatives and lower value are removed, resulting in a more accurate and higher overall value search term package for the target industry.

[0138] In summary, the search term package determination method provided by this invention obtains multiple search terms and multiple recall results corresponding to these search terms. Then, based on these recall results, a search index is determined to identify the search relevance value between each search term and the remaining first search terms among the multiple search terms. Next, industry terms representing the target industry are identified from the multiple search terms. Then, based on the search index of the industry terms and the remaining second search terms among the multiple search terms, a search term package for the target industry is determined. Specifically, by analyzing multiple recall results of multiple search terms containing industry terms in practical applications, a search index is quantified to identify the true search value of each search term in practical applications compared to other search terms. Based on this, for an industry term among multiple search terms, its true search value compared to other search terms can be determined using this search index, thereby identifying the search term package for the corresponding industry among the multiple search terms. Thus, the automatic generation of industry search term packages can be achieved without manual intervention, greatly reducing the generation cost of industry term packages and improving the generation efficiency of search term packages. At the same time, by using real search terms and recall results, the generated search term package is more closely related to the user's real usage scenarios, giving the generated search term package higher search value and accuracy.

[0139] See Figure 7 , Figure 7 This is a flowchart illustrating another method for determining search term packages provided in an embodiment of this application. The method is applied to a search term package determining device and may include the following steps:

[0140] 701: Retrieves multiple search terms and multiple recall results corresponding to the multiple search terms.

[0141] 702: Based on multiple recall results, determine the search metrics for each search term.

[0142] 703: Identify industry terms used to characterize the target industry from multiple search terms.

[0143] 704: Determine the target industry's search term package based on search metrics of industry terms and secondary search terms.

[0144] 705: Retrieves multiple brand terms from the search term package.

[0145] In this implementation, "brand term" refers to the names of various brands in the industry corresponding to the search term package.

[0146] 706: Based on the search popularity of each brand term, multiple brand terms are sorted to obtain the brand ranking of the industry corresponding to the search term package.

[0147] In this embodiment, such as Figure 8 As shown, when a user needs to view brand rankings within an industry, they can send a request to the search term packet determination device via their user terminal. After obtaining the search term packet corresponding to the industry, the search term packet determination device identifies the brand terms for each brand within the packet and sorts these brand terms according to their search popularity from highest to lowest, thus obtaining the brand rankings for the industry. Finally, the search term packet determination device sends the brand rankings to the user terminal for viewing.

[0148] It should be noted that the relevant principles of steps 701-704 can be referred to the relevant descriptions of the above implementation methods, and can solve the same technical problems and achieve the same technical effects, so they will not be repeated here.

[0149] As can be seen, by extracting brand terms used in searches within a defined industry during the current period from a specific industry's search term package, and ranking brands based on the search popularity of each brand term, the system automatically determines brand rankings for the current industry using real search data. Furthermore, the search popularity reveals the true intent of users during searches, thereby making the generated brand rankings more authentic and authoritative.

[0150] See Figure 9 , Figure 9 This is a flowchart illustrating another method for determining search term packages provided in an embodiment of this application. The method is applied to a search term package determining device and may include the following steps:

[0151] 901: Retrieves multiple search terms and multiple recall results corresponding to those search terms.

[0152] 902: Based on multiple recall results, determine the search metrics for each search term.

[0153] 903: Identify industry terms used to characterize the target industry from multiple search terms.

[0154] 904: Determine the target industry's keyword package based on search metrics of industry keywords and secondary search terms.

[0155] 905: Get the number of searches for each search term in the search term package.

[0156] 906: Based on the number of searches for each search term, determine the search trends of the corresponding industry for the search term package.

[0157] In this implementation, the total search volume for each search term in the industry's search term package can be summed to obtain the total search volume for the industry. This total search volume is then combined with the total search volume for other industries to determine the industry's search share within the current period. Finally, a comparative analysis is performed with the industry's search share in historical periods to determine the industry's search trend.

[0158] like Figure 10 As shown, industry search trends can be displayed by drawing line charts or bar charts. Specifically, in this chart, the horizontal axis represents different periods, and the vertical axis represents the search share. Therefore, industry search trends can be presented more intuitively through visual means.

[0159] It should be noted that the relevant principles of steps 901-904 can be referred to the relevant descriptions of the above implementation methods, and can solve the same technical problems and achieve the same technical effects, so they will not be repeated here.

[0160] As can be seen, by statistically analyzing the search frequency of each search term within an industry's search term package over a period, the total number of searches for that industry within that period can be determined. This allows for further analysis of the industry's search share within that period, thus revealing search trends. Compared to traditional search trend analysis methods, this significantly reduces the computational cost of acquiring industry traffic trends, achieving comprehensive pre-calculation coverage across all industries without manual intervention. Furthermore, the search term package data is based on real-world search data, making the calculation of industry search trends more realistic and accurate.

[0161] See Figure 11 , Figure 11 This is a block diagram illustrating the functional modules of a search term packet determination device provided for embodiments of this application. For example... Figure 11 As shown, the search term packet determination device 1100 includes:

[0162] The acquisition module 1101 is used to acquire multiple search terms and multiple recall results corresponding to the multiple search terms;

[0163] Analysis module 1102 is used to determine the search metric for each search term based on multiple recall results. The search metric is a value indicator of the search relevance between each search term and a first search term. The first search term is the search term remaining after removing each search term from the multiple search terms.

[0164] The determination module 1103 is used to determine industry terms that characterize the target industry from multiple search terms;

[0165] The processing module 1104 is used to determine the search term package of the target industry based on the search metrics of industry terms and second search terms, wherein the second search term is the remaining search term after removing industry terms from multiple search terms.

[0166] In an embodiment of the present invention, the search metric for each search term includes the correlation coefficient between each search term and the remaining first search terms, as well as the search popularity of each search term. Based on this, in determining the search metric for each search term based on multiple recall results, the analysis module 1102 is specifically used for:

[0167] Based on the access records of multiple recall results, determine the correlation coefficient between any two search terms;

[0168] The search popularity of each search term is determined based on the correlation coefficient between each search term and the first search term.

[0169] In an embodiment of the present invention, in determining the correlation coefficient between any two search terms based on access records from multiple recall results, the analysis module 1102 is specifically used for:

[0170] Based on the access records of multiple recall results, determine the total number of accesses for each recall result and the number of associated accesses between each recall result and the search term corresponding to that recall result. The associated access count is the number of times the recall result is accessed when the recall result is searched using the corresponding search term.

[0171] The correlation coefficient between any two search terms is determined based on the total number of visits for each recall result and the number of associated visits between each recall result and the search term corresponding to that recall result.

[0172] In an embodiment of the present invention, in determining the search term package for the target industry based on the search metrics of industry terms and second search terms, the processing module 1104 is specifically used for:

[0173] Based on the correlation coefficient between industry terms and second search terms, and the search popularity of industry terms and second search terms, multiple industry search terms are identified in the second search terms. Among them, the correlation coefficient between each industry search term and the industry term is greater than the first threshold, and the ratio of the search popularity of the industry term to the search popularity of each industry search term is greater than the second threshold.

[0174] The industry term and multiple industry search terms are used as a search term package for the corresponding industry.

[0175] In an embodiment of the present invention, regarding the determination of industry terms for characterizing a target industry from multiple search terms, the determination module 1103 is specifically used for:

[0176] Multiple search terms are matched with a preset industry dictionary to determine industry terms, which contains a number of pre-recorded industry terms.

[0177] In an embodiment of the present invention, before matching multiple search terms with a preset industry dictionary to determine industry terms, the determining module 1103 is further configured to:

[0178] Retrieve multiple indivisible minimum business categories from the business category table;

[0179] Use the product name of each minimum business category as the industry term for the corresponding industry of each minimum business category;

[0180] Generate an industry dictionary based on multiple industry terms.

[0181] In an embodiment of the present invention, multiple recall results are data that has been searched and effectively accessed through any one of multiple search terms. Valid access refers to the access time for each recall result being greater than the valid access time threshold corresponding to the data type of that recall result.

[0182] In an embodiment of the present invention, the processing module 1104 is further configured to:

[0183] Obtain historical search term packages for the target industry;

[0184] Based on the search term package and the historical search term package, multiple first-differentiating words, multiple second-differentiating words, and multiple common words are identified. Each first-differentiating word is a word that exists in the search term package but not in the historical term package, and each second-differentiating word is a word that exists in the historical search term package but not in the search term package.

[0185] By filtering the first and second difference words, multiple target words are obtained;

[0186] Use multiple target terms and multiple common terms as the industry-specific search term package.

[0187] In an embodiment of the present invention, in filtering the first and second difference words to obtain multiple target words, the processing module 1104 is specifically used for:

[0188] Based on the similarity between each first difference word and each second difference word, multiple difference word groups are determined among multiple first difference words and multiple second difference words, wherein each difference word group includes a third difference word and a fourth difference word, multiple first difference words include a third difference word, multiple second difference words include a fourth difference word, and the similarity between the third difference word and the fourth difference word is greater than a third threshold.

[0189] Each difference phrase is filtered to obtain multiple fifth difference words;

[0190] The remaining difference words after removing the difference words from multiple difference word groups from multiple fifth difference words, multiple first difference words, and multiple second difference words are taken as multiple target words.

[0191] In an embodiment of the present invention, in filtering each difference word group to obtain multiple fifth difference words, the processing module 1104 is specifically used for:

[0192] Determine the usage popularity of the third and fourth differentiating words in each differentiating phrase within a preset time period;

[0193] If the usage popularity of the fourth difference word is greater than or equal to that of the third difference word, the third and fourth difference words will be identified as the retained words for each difference word group.

[0194] If the usage popularity of the fourth difference word is less than that of the third difference word, determine the difference between the usage popularity of the third difference word and the usage popularity of the fourth difference word;

[0195] If the difference between the usage popularity of the third differentiating word and the usage popularity of the fourth differentiating word is greater than the fourth threshold, the third differentiating word will be determined as a retained word for each differentiating word group.

[0196] If the difference between the usage popularity of the third difference word and the usage popularity of the fourth difference word is less than or equal to the fourth threshold, the third difference word and the fourth difference word are determined as the retained words for each difference word group;

[0197] The retained words corresponding to multiple different word groups are treated as multiple fifth different words.

[0198] In an embodiment of the present invention, the processing module 1104 is further configured to:

[0199] Retrieve multiple brand terms from the search term package;

[0200] Based on the search popularity of each brand term, multiple brand terms are sorted to obtain the brand ranking of the industry corresponding to the search term package.

[0201] In an embodiment of the present invention, the processing module 1104 is further configured to:

[0202] Get the search count for each search term in the search term package;

[0203] Based on the number of searches for each search term, determine the search trends for the corresponding industry for the search term package.

[0204] See Figure 12 , Figure 12 This is a schematic diagram of the structure of an electronic device provided for an embodiment of this application. For example... Figure 12As shown, the electronic device 1200 includes a transceiver 1201, a processor 1202, and a memory 1203. They are connected to each other via a bus 1204. The memory 1203 is used to store computer programs and data, and can transfer the data stored in the memory 1203 to the processor 1202.

[0205] Processor 1202 is used to read the computer program in memory 1203 and perform the following operations:

[0206] Retrieve multiple search terms and multiple recall results corresponding to the multiple search terms;

[0207] Based on multiple recall results, the search metrics for each search term are determined. The search metrics are the value indicators of the search relevance between each search term and the first search term, which is the search term remaining after removing each of the multiple search terms.

[0208] Identify industry terms used to characterize the target industry from multiple search terms;

[0209] Based on the search metrics of industry terms and secondary search terms, a search term package for the target industry is determined. The secondary search term is the search term remaining after removing the industry terms from a pool of search terms.

[0210] In an embodiment of the present invention, the search metric for each search term includes the correlation coefficient between each search term and the remaining first search terms, as well as the search popularity of each search term. Based on this, in determining the search metric for each search term based on multiple recall results, the processor 1202 is specifically configured to perform the following operations:

[0211] Based on the access records of multiple recall results, determine the correlation coefficient between any two search terms;

[0212] The search popularity of each search term is determined based on the correlation coefficient between each search term and the first search term.

[0213] In an embodiment of the present invention, in determining the correlation coefficient between any two search terms based on access records of multiple recall results, the processor 1202 is specifically configured to perform the following operations:

[0214] Based on the access records of multiple recall results, determine the total number of accesses for each recall result and the number of associated accesses between each recall result and the search term corresponding to that recall result. The associated access count is the number of times the recall result is accessed when the recall result is searched using the corresponding search term.

[0215] The correlation coefficient between any two search terms is determined based on the total number of visits for each recall result and the number of associated visits between each recall result and the search term corresponding to that recall result.

[0216] In an embodiment of the present invention, in determining a search term package for a target industry based on search metrics of industry terms and second search terms, the processor 1202 is specifically configured to perform the following operations:

[0217] Based on the correlation coefficient between industry terms and second search terms, and the search popularity of industry terms and second search terms, multiple industry search terms are identified in the second search terms. Among them, the correlation coefficient between each industry search term and the industry term is greater than the first threshold, and the ratio of the search popularity of the industry term to the search popularity of each industry search term is greater than the second threshold.

[0218] The industry term and multiple industry search terms are used as a search term package for the corresponding industry.

[0219] In an embodiment of the present invention, regarding determining industry terms for characterizing a target industry from a plurality of search terms, processor 1202 is specifically configured to perform the following operations:

[0220] Multiple search terms are matched with a preset industry dictionary to determine industry terms, which contains a number of pre-recorded industry terms.

[0221] In an embodiment of the present invention, before matching multiple search terms with a preset industry dictionary to determine industry terms, the processor 1202 is further configured to perform the following operations:

[0222] Retrieve multiple indivisible minimum business categories from the business category table;

[0223] Use the product name of each minimum business category as the industry term for the corresponding industry of each minimum business category;

[0224] Generate an industry dictionary based on multiple industry terms.

[0225] In an embodiment of the present invention, multiple recall results are data that has been searched and effectively accessed through any one of multiple search terms. Valid access refers to the access time for each recall result being greater than the valid access time threshold corresponding to the data type of that recall result.

[0226] In an embodiment of the present invention, the processor 1202 is further configured to perform the following operations:

[0227] Obtain historical search term packages for the target industry;

[0228] Based on the search term package and the historical search term package, multiple first-differentiating words, multiple second-differentiating words, and multiple common words are identified. Each first-differentiating word is a word that exists in the search term package but not in the historical term package, and each second-differentiating word is a word that exists in the historical search term package but not in the search term package.

[0229] By filtering the first and second difference words, multiple target words are obtained;

[0230] Use multiple target terms and multiple common terms as the industry-specific search term package.

[0231] In an embodiment of the present invention, in filtering the first and second difference words to obtain multiple target words, the processor 1202 is specifically configured to perform the following operations:

[0232] Based on the similarity between each first difference word and each second difference word, multiple difference word groups are determined among multiple first difference words and multiple second difference words, wherein each difference word group includes a third difference word and a fourth difference word, multiple first difference words include a third difference word, multiple second difference words include a fourth difference word, and the similarity between the third difference word and the fourth difference word is greater than a third threshold.

[0233] Each difference phrase is filtered to obtain multiple fifth difference words;

[0234] The remaining difference words after removing the difference words from multiple difference word groups from multiple fifth difference words, multiple first difference words, and multiple second difference words are taken as multiple target words.

[0235] In an embodiment of the present invention, in filtering each difference word group to obtain multiple fifth difference words, the processor 1202 is specifically configured to perform the following operations:

[0236] Determine the usage popularity of the third and fourth differentiating words in each differentiating phrase within a preset time period;

[0237] If the usage popularity of the fourth difference word is greater than or equal to that of the third difference word, the third and fourth difference words will be identified as the retained words for each difference word group.

[0238] If the usage popularity of the fourth difference word is less than that of the third difference word, determine the difference between the usage popularity of the third difference word and the usage popularity of the fourth difference word;

[0239] If the difference between the usage popularity of the third differentiating word and the usage popularity of the fourth differentiating word is greater than the fourth threshold, the third differentiating word will be determined as a retained word for each differentiating word group.

[0240] If the difference between the usage popularity of the third difference word and the usage popularity of the fourth difference word is less than or equal to the fourth threshold, the third difference word and the fourth difference word are determined as the retained words for each difference word group;

[0241] The retained words corresponding to multiple different word groups are treated as multiple fifth different words.

[0242] In an embodiment of the present invention, the processor 1202 is further configured to perform the following operations:

[0243] Retrieve multiple brand terms from the search term package;

[0244] Based on the search popularity of each brand term, multiple brand terms are sorted to obtain the brand ranking of the industry corresponding to the search term package.

[0245] In an embodiment of the present invention, the processor 1202 is further configured to perform the following operations:

[0246] Get the search count for each search term in the search term package;

[0247] Based on the number of searches for each search term, determine the search trends for the corresponding industry for the search term package.

[0248] It should be understood that the search term determination device in this application may include smartphones (such as Android phones, iOS phones, Windows Phones, etc.), tablet computers, PDAs, laptops, mobile internet devices (MIDs), robots, or wearable devices, etc. The above-mentioned search term determination devices are merely examples, not exhaustive, and include, but are not limited to, the search term determination devices described above. In practical applications, the above-mentioned search term determination device may also include: intelligent vehicle terminals, computer equipment, etc.

[0249] Through the above description of the embodiments, those skilled in the art can clearly understand that the present invention can be implemented by means of software combined with a hardware platform. Based on this understanding, all or part of the technical solution of the present invention that contributes to the background art can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of the present invention.

[0250] Therefore, embodiments of this application also provide a computer-readable storage medium storing a computer program that is executed by a processor to implement some or all of the steps of any of the search term package determination methods described in the above method embodiments. For example, the storage medium may include a hard disk, floppy disk, optical disk, magnetic tape, magnetic disk, USB flash drive, flash memory, etc.

[0251] This application also provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program operable to cause a computer to perform some or all of the steps of any of the search term package determination methods described in the above method embodiments.

[0252] It should be noted that, for the sake of simplicity, the aforementioned methods are described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are optional, and the actions and modules involved are not necessarily essential to this application.

[0253] In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0254] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical or other forms.

[0255] 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 units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0256] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software program module.

[0257] If the integrated unit is implemented as a software program module and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0258] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage device, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0259] The embodiments of this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The above description of the embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for determining a search term package, characterized in that, The method includes: Obtain multiple search terms and multiple recall results corresponding to the multiple search terms; Based on the multiple recall results, a search metric is determined for each search term. The search metric is a value indicator of the search relevance between each search term and a first search term. The first search term is the remaining search term after removing each of the multiple search terms. The search metric for each search term includes the correlation coefficient between each search term and the first search term, and the search popularity of each search term; including: Based on the access records of the multiple recall results, determine the total number of accesses for each recall result and the number of associated accesses between each recall result and the search term corresponding to the recall result, wherein the number of associated accesses is the number of times the recall result is accessed when the recall result is searched through the corresponding search term; Based on the total number of visits to each recall result and the number of associated visits between each recall result and the search term corresponding to that recall result, the correlation coefficient between any two search terms is determined; based on the correlation coefficient between each search term and the first search term, the search popularity of each search term is determined. Identify industry terms from the plurality of search terms to characterize the target industry; Based on the search metrics of the industry term and the second search term, a search term package for the target industry is determined, wherein the second search term is the remaining search term after removing the industry term from the plurality of search terms.

2. The method according to claim 1, characterized in that, The step of determining the search term package for the target industry based on the search metrics of the industry term and the second search term includes: Based on the correlation coefficient between the industry term and the second search term, and the search popularity of the industry term and the second search term, multiple industry search terms are determined in the second search term, wherein the correlation coefficient between each industry search term and the industry term is greater than a first threshold, and the ratio of the search popularity of the industry term to the search popularity of each industry search term is greater than a second threshold. The industry term and the multiple industry search terms are used as the search term package for the industry corresponding to the industry term.

3. The method according to claim 1 or 2, characterized in that, The process of determining industry terms to characterize the target industry from the plurality of search terms includes: The multiple search terms are matched with a preset industry dictionary to determine the industry terms, wherein the industry dictionary contains multiple industry terms in advance.

4. The method according to claim 3, characterized in that, Before matching the multiple search terms with a preset industry dictionary to determine the industry term, the method further includes: Retrieve multiple indivisible minimum business categories from the business category table; The product name of each minimum business category is used as the industry term for the corresponding industry of each minimum business category; The industry dictionary is generated based on multiple industry terms.

5. The method according to claim 1, characterized in that, The multiple recall results are data that has been searched and effectively accessed through any one of the multiple search terms. The effective access refers to the access time for each recall result being greater than the effective access time threshold corresponding to the data type of the recall result.

6. The method according to claim 1, characterized in that, The method further includes: Obtain the historical search term package for the target industry; Based on the search term package and the historical search term package, multiple first difference words, multiple second difference words, and multiple common words are determined, wherein each first difference word is a word that exists in the search term package but not in the historical term package, and each second difference word is a word that exists in the historical search term package but not in the search term package; By filtering the first and second difference words, multiple target words are obtained; The multiple target words and the multiple common words are used as the industry-specific search term package corresponding to the search term package.

7. The method according to claim 6, characterized in that, The process of filtering the first and second difference words yields multiple target words, including: Based on the similarity between each first difference word and each second difference word, multiple difference word groups are determined among the multiple first difference words and the multiple second difference words, wherein each difference word group includes a third difference word and a fourth difference word, the multiple first difference words include the third difference word, the multiple second difference words include the fourth difference word, and the similarity between the third difference word and the fourth difference word is greater than a third threshold. Each of the different word groups is filtered to obtain multiple fifth different words; The remaining difference words after removing the difference words in the multiple difference word groups from the multiple fifth difference words, the multiple first difference words, and the multiple second difference words are taken as the multiple target words.

8. The method according to claim 7, characterized in that, The process of filtering each of the different word groups yields multiple fifth different words, including: Determine the usage popularity of the third and fourth differential words in each differential word group within a preset time period; If the usage popularity of the fourth difference word is greater than or equal to the usage popularity of the third difference word, the third difference word and the fourth difference word are determined as the retained words of each difference word group; If the usage popularity of the fourth differentiating word is less than that of the third differentiating word, the difference between the usage popularity of the third differentiating word and the usage popularity of the fourth differentiating word is determined. If the difference between the usage popularity of the third difference word and the usage popularity of the fourth difference word is greater than the fourth threshold, the third difference word is determined as the retained word of each difference word group; If the difference between the usage popularity of the third difference word and the usage popularity of the fourth difference word is less than or equal to the fourth threshold, the third difference word and the fourth difference word are determined as the retained words of each difference word group; The retained words corresponding to the plurality of differential word groups are regarded as the plurality of fifth differential words.

9. The method according to claim 1, characterized in that, The method further includes: Retrieve multiple brand terms from the search term package; Based on the search popularity of each brand term, the multiple brand terms are sorted to obtain the brand ranking of the industry corresponding to the search term package.

10. The method according to claim 1, characterized in that, The method further includes: Obtain the search count for each search term in the search term package; Based on the number of searches for each search term, the search trends of the industries corresponding to the search term package are determined.

11. A search term packet determination device, characterized in that, The device includes: The acquisition module is used to acquire multiple search terms and multiple recall results corresponding to the multiple search terms; An analysis module is used to determine the search metric for each search term based on the multiple recall results. The search metric is a value indicator of the search relevance between each search term and a first search term. The first search term is the remaining search term after removing each of the multiple search terms. The search metric for each search term includes the correlation coefficient between each search term and the first search term, and the search popularity of each search term; including: Based on the access records of the multiple recall results, determine the total number of accesses for each recall result and the number of associated accesses between each recall result and the search term corresponding to the recall result, wherein the number of associated accesses is the number of times the recall result is accessed when the recall result is searched through the corresponding search term; Based on the total number of visits to each recall result and the number of associated visits between each recall result and the search term corresponding to that recall result, the correlation coefficient between any two search terms is determined; based on the correlation coefficient between each search term and the first search term, the search popularity of each search term is determined. The determination module is used to determine industry terms that characterize the target industry from the plurality of search terms; The processing module is used to determine the search term package of the target industry based on the search metrics of the industry term and the second search term, wherein the second search term is the remaining search term after removing the industry term from the plurality of search terms.

12. An electronic device, characterized in that, The method includes a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the one or more programs include instructions for performing the steps of the method according to any one of claims 1-10.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that is executed by a processor to implement the method as described in any one of claims 1-10.

Images