Candidate word association method, apparatus, and program product

Abstract

This application relates to a candidate word association method, apparatus, and program product, belonging to the field of computer technology, which can improve the accuracy of candidate word association. The method includes: responding to an input operation by a user, determining input information, and obtaining the page resource object category of the page browsed by the user; the page resource object category is determined based on the category to which the resource objects displayed on the page belong; using the input information as a candidate word prefix, determining multiple first resource object candidate words containing the candidate word prefix, and determining the correlation between each first resource object candidate word and the page resource object category; based on the correlation, determining at least one second resource object candidate word from the multiple first resource object candidate words; obtaining and displaying the candidate word association result of the input information based on the second resource object candidate word.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus, computer device, computer-readable storage medium, and computer program product. Background Technology

[0002] With the development of computer technology, when a user performs an input operation, multiple related candidate words can be provided to the user through candidate word association, reducing the number of characters the user needs to continue typing.

[0003] In related technologies, the system typically has a pre-built dictionary containing common nouns. When a user inputs a string, it returns all candidate words that are prefixed with that string. For example, if the user inputs "fragrant," the dictionary can match "perfume," "sausage," "aromatherapy," etc. This type of method is simple to implement and responds quickly.

[0004] However, the candidate words provided in this way may not match the user's current context. For example, if a user enters "fragrance" on an e-commerce page showcasing high-end perfumes, the system may still prioritize returning "sausage" or "sesame oil" from the dictionary due to word frequency or dictionary order. However, the returned candidate words are severely out of touch with the current page context, and the accuracy of the candidate word association still needs improvement. Summary of the Invention

[0005] Therefore, it is necessary to provide a candidate word association method, apparatus, computer device, computer-readable storage medium, and computer program product to address the aforementioned technical problems.

[0006] Firstly, this application provides a candidate word association method, including:

[0007] In response to an input operation by a user object, the input information is determined, and the page resource object category of the page browsed by the user object is obtained; the page resource object category is determined according to the category to which the resource objects displayed on the page belong.

[0008] Using the input information as a candidate word prefix, determine multiple first resource object candidate words containing the candidate word prefix, and determine the correlation between each first resource object candidate word and the page resource object category;

[0009] Based on the relevance, at least one second resource object candidate word is determined from the plurality of first resource object candidate words;

[0010] The candidate word association results of the input information are obtained and displayed based on the candidate words of the second resource object.

[0011] In one embodiment, determining the correlation between each candidate word of the first resource object and the category of the page resource object includes:

[0012] Obtain a pre-built association mapping table; the association mapping table records the association degree between multiple resource object descriptors and the page resource object categories;

[0013] Based on the matching results between each candidate word of the first resource object and the relevance mapping table, the relevance between each candidate word of the first resource object and the category of the page resource object is determined.

[0014] In one embodiment, the correlation mapping table is constructed through the following steps:

[0015] Obtain historical search data for each of multiple sample user objects; the historical search data includes historical pages corresponding to the page resource object categories in which the sample user objects have performed input operations, and resource object descriptors corresponding to the resource objects selected by the sample user objects based on the historical pages;

[0016] Based on the statistical results of the historical search data, the co-occurrence frequency of the page resource object category and each resource object descriptor is determined, and the correlation between each resource object descriptor and the page resource object category is determined based on the co-occurrence frequency.

[0017] A correlation mapping table is constructed based on the correlation degree of each of the resource object descriptors.

[0018] In one embodiment, determining the relevance between each candidate term of the first resource object and the category of the page resource object based on the matching result of each candidate term of the first resource object and the relevance mapping table includes:

[0019] If the correlation mapping table contains resource object descriptive words that match the first resource object candidate words, then the correlation between the first resource object candidate words and the page resource object category is determined based on the correlation of the matching resource object descriptive words in the correlation mapping table.

[0020] If there is no resource object description word matching the first resource object candidate word in the correlation mapping table, the correlation between the first resource object candidate word and the page resource object category is determined according to a preset value less than the correlation threshold.

[0021] In one embodiment, obtaining the page resource object category of the page browsed by the user object includes:

[0022] The page browsed by the user object is determined, and the category identification information of the page is extracted; the category identification information includes at least one of page attribute information and page content;

[0023] The page resource object category of the page is determined based on the category identification information.

[0024] In one embodiment, determining a plurality of first resource object candidate words containing the candidate word prefix, using the input information as a candidate word prefix, includes:

[0025] Obtain the prefix tree; the prefix tree includes multiple resource object candidate words;

[0026] Using the input information as a prefix for candidate words, a search is performed in the prefix tree, and multiple candidate words for first resource objects are obtained based on the search results.

[0027] In one embodiment, obtaining the page resource object category of the page browsed by the user object includes:

[0028] Determine the page resource object category of the first page currently being viewed by the user object; and / or,

[0029] Determine the page resource object category of the second page previously viewed by the user object; the second page is different from the first page currently viewed by the user object, and the input operation is triggered on the first page.

[0030] In one embodiment, determining at least one second resource object candidate word from the plurality of first resource object candidate words based on the relevance includes:

[0031] Obtain the search popularity of each candidate word for the first resource object;

[0032] For each candidate word of the first resource object, the recommendation degree of the candidate word of the first resource object is determined based on the relevance and search popularity of the candidate word of the first resource object.

[0033] At least one of the first resource object candidate words whose recommendation level meets the preset recommendation level conditions is determined as the second resource object candidate word.

[0034] Secondly, this application also provides a candidate word association device, comprising:

[0035] The category recognition module is used to respond to the input operation of the user object, determine the input information, and obtain the page resource object category of the page browsed by the user object; the page resource object category is determined according to the category to which the resource object displayed on the page belongs;

[0036] The relevance determination module is used to determine multiple first resource object candidate words containing the candidate word prefix by using the input information as a candidate word prefix, and to determine the relevance between each first resource object candidate word and the page resource object category;

[0037] The candidate word filtering module is used to determine at least one second resource object candidate word from the plurality of first resource object candidate words based on the relevance.

[0038] The association result determination module is used to obtain and display the candidate word association results of the input information based on the candidate words of the second resource object.

[0039] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the candidate word association method as described above.

[0040] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the candidate word association method as described above.

[0041] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the candidate word association method as described above.

[0042] The aforementioned candidate word association method, apparatus, computer device, computer-readable storage medium, and computer program product, in response to a user's input operation, can determine the input information and obtain the page resource object category of the page browsed by the user, wherein the page resource object category is determined according to the category to which the resource objects displayed on the page belong; then, using the input information as a candidate word prefix, multiple first resource object candidate words containing the candidate word prefix are determined, and the correlation degree between each first resource object candidate word and the page resource object category is determined; furthermore, based on the correlation degree, at least one second resource object candidate word is determined from the multiple first resource object candidate words, and the candidate word association result of the input information is obtained and displayed based on the second resource object candidate word. In this embodiment, by identifying the page resource object category of the page browsed by the user, the correlation between each first resource object candidate word and the page resource object category is determined, and based on the correlation, second resource object candidate words are determined from multiple first resource object candidate words. This can effectively combine the user's actual resource object browsing scenario and accurately predict the potential needs of the user's current input operation. Compared with the association method based solely on input information, it significantly improves the matching degree between the candidate word association results and the current input scenario, helps to improve the accuracy of candidate word association, and reduces the display of irrelevant candidate words. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1a This is a schematic diagram of one type of candidate word association results;

[0045] Figure 1b This is a schematic diagram of the association results for another candidate word;

[0046] Figure 2 This is a flowchart illustrating the candidate word association method in one embodiment;

[0047] Figure 3 This is a flowchart illustrating a step in one embodiment of determining the correlation between candidate words of a first resource object and resource categories of a page object;

[0048] Figure 4 This is a flowchart illustrating another step in one embodiment for determining the correlation between candidate words of a first resource object and resource categories of a page object;

[0049] Figure 5 This is a flowchart illustrating the candidate word association method in another embodiment;

[0050] Figure 6 This is a structural block diagram of a candidate word association device in one embodiment;

[0051] Figure 7 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0053] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various objects, but these objects are not limited by these terms. These terms are only used to distinguish the first object from the second object. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the solutions, or any combination of multiple solutions.

[0054] It should be noted that the user information (including but not limited to user device information, user personal information, user browsing records, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.

[0055] With the development of computer technology, when a user performs an input operation, multiple relevant candidate words can be provided for the user through candidate word association, reducing the characters that the user needs to continue to input subsequently.

[0056] In related technologies, the system usually pre-sets a word library, which contains common nouns. When the user inputs a string, all candidate words starting with this string are returned. For example, if the user inputs "fragrance", "perfume", "sausage", "aromatherapy", etc. can be matched through the word library. Such a method is simple to implement and has a rapid response.

[0057] However, this method completely relies on a static word library and string matching, and cannot perceive the application scenario or interaction context where the user is currently located. The candidate words provided by the system may not match the current scenario of the user, that is, the same candidate word association results will be returned in different scenarios. For example, when the user inputs "fragrance" on an e-commerce page displaying high-end perfumes, the system may still preferentially return "sausage" or "sesame oil" in the word library due to word frequency or lexicographical order reasons. Another example is when Figure 1a the user inputs "fragrance" on the home page of an e-commerce platform as shown in Figure 1b and when the user inputs "fragrance" on the personal care and cleaning product category page of the same e-commerce platform as shown in

[0058]

[0059] This application provides a candidate word association method, device, computer device, computer-readable storage medium, and computer program product, which can effectively enhance the relevance between the candidate word association result and the scenario where the user is located, and improve the accuracy of candidate word association.

[0060] In one embodiment, a candidate word association method is provided. This embodiment illustrates the application of this method to a terminal, but it is understood that the method can also be applied to a server, or to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. The terminal can be, but is not limited to, various personal computers, laptops, smartphones, tablets, drones, low-altitude aircraft, IoT devices, and portable wearable devices. IoT devices can be smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. Portable wearable devices can be smartwatches, smart bracelets, head-mounted devices, etc. Head-mounted devices can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0061] like Figure 2 As shown, in this embodiment, the method may include the following steps:

[0062] S201, in response to the user object's input operation, determine the input information and obtain the page resource object category of the page browsed by the user object; the page resource object category is determined according to the category to which the resource object displayed on the page belongs.

[0063] The user object can be any user who performs the input operation.

[0064] Input operations can be one or more operations that trigger information input, and for example, they can include one or more methods such as keyboard input and voice input.

[0065] Input information can refer to text obtained from input operations based on user objects.

[0066] Page resource object category can be understood as the category of resource objects displayed on the browsing page, such as food, appliances, etc. The page resource object category can be determined by the category to which the resources on the page belong.

[0067] Resource objects can be various resources provided to user objects. For example, resource objects can include one or more of physical resources, virtual resources, and service resources.

[0068] In the specific implementation, when a user object triggers an input operation, the information entered by the user object through the input operation can be obtained accordingly, i.e., the input information.

[0069] On the other hand, it is also possible to obtain and analyze the pages viewed by the user, and determine the category of the resource objects displayed on the page based on the analysis results. In some possible embodiments, when determining which pages viewed by the user to analyze, the pages can be filtered based on the browsing time or the page type, and then the filtered pages are analyzed to determine their page resource object types. For example, recently viewed pages that meet the browsing time requirement can be selected as the pages to be analyzed, or pages of a specified type can be selected as the pages to be analyzed, such as browser pages, in-application pages, game pages, or one or more of these.

[0070] S202, using the input information as a candidate word prefix, determine multiple first resource object candidate words containing the candidate word prefix, and determine the correlation between each first resource object candidate word and the page resource object category.

[0071] The resource object descriptor can be a word or phrase used to describe the characteristics of a resource object. In some exemplary embodiments, the resource object descriptor may include the category name of the resource object, such as "perfume," or the brand of the resource object, such as "XX brand." The first resource object candidate word can be understood as a resource object descriptor word generated based on the input information that may meet the user's needs. The relevance can characterize the degree of correlation between the first resource object candidate word and the page resource object type in terms of category affiliation and / or. In some examples, it can be calculated through a preset vocabulary association database or semantic analysis algorithm.

[0072] In practice, after obtaining the input information, it can be used as a prefix for candidate words to determine multiple candidate words for the first resource object. The specific acquisition method can be selected according to the actual situation. For example, multiple candidate words for the first resource object can be determined based on the matching result between the input information and the prefix tree, or multiple extended words can be obtained by expanding the input information, and then multiple candidate words for the first resource object can be determined based on the recommendation degree of the multiple extended words.

[0073] After obtaining multiple candidate words for the first resource object, for each candidate word, the correlation between the candidate word and the category of the page resource object can be obtained.

[0074] S203, Based on the relevance, determine at least one second resource object candidate word from multiple first resource object candidate words.

[0075] After obtaining the associations of each candidate term for the first resource object, multiple candidate terms for the first resource object can be filtered based on their association degree, and candidate terms for the second resource object can be obtained based on the filtering results. The candidate terms for the second resource object can include words selected from the candidate terms for the first resource object that have a high degree of association with the page resource object type.

[0076] In some possible embodiments, if candidate words for the first resource object are filtered according to relevance, the filtering criteria can be set as a relevance threshold or a relevance ranking. In other embodiments, in addition to relevance, candidate words for the first resource object can also be filtered by combining information from other dimensions. For example, candidate words for the second resource object can be obtained by filtering based on the user's historical input records of the candidate words for the first resource object (such as input time, number of inputs, or one or more other information) and / or the global usage of the candidate words for the first resource object.

[0077] S204. Obtain and display the candidate word association results of the input information based on the candidate words of the second resource object.

[0078] The candidate word association results refer to the content that presents candidate words for the second resource object to the user in a visual format, facilitating quick selection and reducing input. The candidate word association results may include one or more candidate words for the second resource object.

[0079] In one embodiment, when there are multiple candidate words for the second resource object, the multiple candidate words for the second resource object can be sorted, for example, sorted from high to low relevance, and then a candidate word association list can be generated based on the sorting results and displayed below the input area as the candidate word association results for the input information.

[0080] In the aforementioned candidate word association method, in response to a user's input operation, the input information can be determined, and the page resource object category of the page browsed by the user can be obtained. The page resource object category is determined based on the category to which the resource objects displayed on the page belong. Then, using the input information as a candidate word prefix, multiple first resource object candidate words containing the candidate word prefix are determined, and the correlation between each first resource object candidate word and the page resource object category is determined. Furthermore, based on the correlation, at least one second resource object candidate word is determined from the multiple first resource object candidate words, and the candidate word association result for the input information is obtained and displayed based on the second resource object candidate word. In this embodiment, by identifying the page resource object category of the page browsed by the user, determining the correlation between each first resource object candidate word and the page resource object category, and determining the second resource object candidate word from the multiple first resource object candidate words based on the correlation, the method can effectively combine the user's actual resource object browsing scenario to accurately predict the potential needs of the user's current input operation. Compared to association methods based solely on input information, this significantly improves the matching degree between the candidate word association result and the current input scenario, thus helping to improve the accuracy of candidate word association.

[0081] In one embodiment, such as Figure 3 As shown, in step S102, determining the relevance between each candidate word of the first resource object and the category of the page resource object may include the following steps:

[0082] S301, obtain the pre-built association mapping table; the association mapping table records the association between multiple resource object descriptors and page resource object categories.

[0083] In this step, the constructed association mapping table can be obtained. The association mapping table can be a pre-constructed data table or data structure used to store the degree of association between resource object descriptors and page resource object categories. It can quantify the relationship between the two, providing data support for subsequent identification of page resource object categories based on descriptors.

[0084] In some possible embodiments, a large number of resource object descriptors under different page resource object categories can be collected in advance to obtain the correlation information between various resource object descriptors under each page resource object category and that category. Then, a correlation mapping table can be constructed based on the correlation information of each resource object descriptor.

[0085] In one embodiment, the relevance mapping table is pre-built offline. When the device first installs the application or performs an application update, a lightweight "category-product name relevance table", i.e., the relevance mapping table, can be downloaded from the server. The file size of this table is small (e.g., the file size is 1M), which can improve the matching of candidate word association results with the user's scenario while reducing the occupation of local storage space.

[0086] For example, the association mapping table may include the following:

[0087] {

[0088] "CATEGORY_PERFUME": {

[0089] "Perfume":1,

[0090] Cologne: 0.92

[0091] Fragrance: 0.85

[0092] "Soap": 0.20,

[0093] "Sausage": 0.05

[0094] },

[0095] "CATEGORY_FOOD": {

[0096] "Sausage": 0.90,

[0097] "Shiitake mushroom": 0.85,

[0098] "Perfume": 0.10

[0099] Chili pepper: 0.05

[0100] }

[0101] }

[0102] The correlation mapping table can be derived based on statistical analysis of historical user behavior data from the platform.

[0103] S302, based on the matching results of each first resource object candidate word and the relevance mapping table, determine the relevance between each first resource object candidate word and the page resource object category.

[0104] After obtaining the relevance mapping table, since the table already records the relevance between each resource object descriptor and the page resource object category, for each first resource object candidate word, by matching the first resource object candidate word with multiple resource object descriptors in the relevance mapping table, the relevance between the first resource object candidate word and the page resource object list can be determined based on the matching results.

[0105] In some possible embodiments, the matching process can employ either exact matching or fuzzy matching. Exact matching directly searches the relevance mapping table for a resource object descriptor that is completely identical to the first resource object candidate word. Fuzzy matching, when no completely identical descriptor exists, searches for descriptors that are semantically similar or partially identical to the candidate word (e.g., by calculating cosine similarity, edit distance, etc.). By combining exact and fuzzy matching, the relevance can be accurately determined even when the first resource object candidate word is not completely identical to the descriptors in the mapping table.

[0106] In this embodiment, by pre-constructing a correlation mapping table, the problem of having to re-analyze and calculate every time the correlation is determined is avoided, which greatly improves the efficiency of determining the correlation between resource object descriptors and page resource object categories.

[0107] In one embodiment, such as Figure 4 As shown, the correlation mapping table can be constructed through the following steps:

[0108] S401, Obtain the historical search data of each of the multiple sample user objects; the historical search data includes the historical pages corresponding to the page resource object categories in which the sample user objects have performed input operations, and the resource object descriptors corresponding to the resource objects selected by the sample user objects based on the historical pages.

[0109] The sample user objects can be a set of user objects selected from the user group. For example, they can be determined by random sampling, stratified sampling, etc.

[0110] In this step, multiple sample user objects can be identified, and their respective historical search data can be obtained. The use of historical search data is with the consent of the sample user objects and has been anonymized. Historical search data characterizes the search behavior of sample user objects through historical pages corresponding to page resource object categories, and the resource object descriptors corresponding to the resource objects that ultimately meet the needs of the sample user objects.

[0111] In this embodiment, historical search data may include historical pages corresponding to a certain page resource object category where the sample user has performed input operations. The sample user can input relevant information on the historical pages, i.e., historical input information. Historical input information may refer to the text content entered by the sample user in the search control of the page corresponding to the page resource object category (i.e., the historical page). In addition, it may also include the specific resource object selected by the sample user through the historical page of the page resource object category, and the resource object descriptor corresponding to the resource object. For example, the sample user enters the prefix P (e.g., fragrance) in the search box or input box of the product page of category C (e.g., category 100203), and finally displays, clicks, or purchases a resource object with the product name W (e.g., the resource object descriptor "light fragrance").

[0112] In this step, the operation data of the sample user object in the historical pages corresponding to different page resource object categories can be obtained, which can be used to form the historical search data of the sample user object. This data can cover the historical pages corresponding to the page resource object categories in which the sample user object actively performed input operations, and the resource object descriptors corresponding to the selected resource object.

[0113] S402, Based on the statistical results of historical search data, determine the co-occurrence frequency of page resource object categories and each resource object descriptor, and determine the correlation between each resource object descriptor and the page resource object category based on the co-occurrence frequency.

[0114] Co-occurrence frequency can refer to the number of times a certain resource object descriptor appears in the historical search data of the same resource object category on the same page. It can be understood as the frequency of the descriptor appearing together with the corresponding category. For example, in the historical search data of the historical page corresponding to category C, the number of times the sample user finally selected resource object W.

[0115] In this step, historical search data can be statistically analyzed to identify the historical pages corresponding to the page resource object categories where sample users performed input operations, as well as the resource object descriptors obtained through searches on those pages. This allows for the calculation of the co-occurrence frequency between the page and a specific resource object descriptor, and the correlation between the resource object descriptor and the page resource object category can be determined based on this co-occurrence frequency. In other words, for a specific resource object descriptor, the correlation between the resource object descriptor and the sample user's search behavior on historical pages of a particular page resource object category can be analyzed based on the statistical results of historical search data and the determined co-occurrence frequency, thereby determining the correlation between the resource object descriptor and that page resource object category. This correlation reflects the tightness of the association between the resource object descriptor and the page resource object category; a higher correlation value indicates a stronger association between the resource object descriptor and the corresponding category.

[0116] For example, after obtaining historical search data from multiple sample user objects across historical pages of category c, multiple (c,w) (or (w,c)) can be identified, where w is a resource object descriptor. For each (c,w), the co-occurrence frequency Count(c,w) can be calculated. Here, Count(c,w) represents the number of times a sample user object ultimately selects w across historical pages of category c. Then, the correlation degree Rel(c,w) between the resource object descriptor w and category c can be determined based on the normalized calculation result of Count(c,w). In one example, the correlation degree Rel(c,w) can be calculated as follows:

[0117] Rel(c,w)=

[0118] In a specific example, on a men's fragrance page, after multiple sample users entered the prefix "fragrance", the co-occurrence frequency of the final selected product name is shown in Table 1 below:

[0119] Table 1

[0120]

[0121] As shown in the table above, maxCount = 9600;

[0122] rel(fragrance, c) = log(1+9600) / log(1+9600) = 1;

[0123] rel(cologne, c) = log(1+8500) / log(1+9600) = 0.987;

[0124] rel(soap,c)=log(1+7000) / log(1+9600)=0.96.

[0125] S403, construct a correlation mapping table based on the correlation degree of each of the multiple resource object descriptors.

[0126] In this step, after obtaining the correlation between each of the multiple resource object descriptors and the page resource object category, they can be organized into a correlation mapping table according to a preset format.

[0127] In this embodiment, on the one hand, historical search data of sample users can be obtained, covering historical pages corresponding to page resource object categories where users have performed input operations, as well as resource object descriptors corresponding to selected resource objects. This effectively reflects the association between the subsequently obtained correlation mapping table and user objects in the corresponding scenarios (such as interacting on historical pages corresponding to page resource object categories). On the other hand, by statistically analyzing co-occurrence frequencies and calculating correlation, the association between user input behavior and resource object categories is transformed into a quantifiable indicator, which can accurately reflect the closeness of the association between resource object descriptors and resource object categories, providing a reliable basis for subsequently screening candidate words for the first resource object.

[0128] In one embodiment, in step S302, determining the relevance between each first resource object candidate word and the page resource object category based on the matching result of the relevance mapping table may include the following steps:

[0129] If the relevance mapping table contains a resource object descriptor that matches the first resource object candidate word, then the relevance between the first resource object candidate word and the page resource object category is determined based on the relevance of the resource object descriptor that matches the first resource object candidate word in the relevance mapping table; if the relevance mapping table does not contain a resource object descriptor that matches the first resource object candidate word, then the relevance between the first resource object candidate word and the page resource object category is determined based on a preset value that is less than the relevance threshold.

[0130] The relevance threshold can be a critical value used to indicate whether there is a valid association between the candidate words of the first resource object and the category of the page resource object. It can be set according to the actual application scenario. If the relevance is higher than this value, it is considered to have a valid association; if it is lower, it is considered to have no valid association. The preset value of less than the relevance threshold refers to a fixed value that is pre-set and lower than the relevance threshold, such as 0.1, which is used to handle the case where there is no matching item in the relevance mapping table.

[0131] When determining the relevance between the first resource object candidate words and the page resource object categories, a matching operation can be performed between the first resource object candidate words and the resource object descriptors in the relevance mapping table. This matching operation can determine whether there are resource object descriptors in the relevance mapping table that match the currently processed first resource object candidate words.

[0132] If, after the matching operation, it is determined that there is a resource object descriptor in the relevance mapping table that matches the candidate first resource object, then the relevance between the candidate first resource object and the page resource object category can be determined directly based on the relevance corresponding to the matching resource object descriptor in the relevance mapping table. If, after the matching operation, it is found that there is no resource object descriptor in the relevance mapping table that matches the candidate first resource object, then in this case, the relevance between the candidate first resource object and the page resource object category can be determined based on a preset value that is less than the relevance threshold.

[0133] For example, suppose the resource object category of the currently extracted page is cat. The relevance mapping table corresponding to cat can be stored in advance on the terminal device. After determining multiple first resource object candidate words, for each first resource object candidate word w, the relevance rel(w,cat) can be obtained by looking up the table when w belongs to the relevance mapping table of category c. If no match is found, the relevance of the first resource object candidate word is set to the default value of 0.1. The default value indicates that the first resource object candidate word is weakly related to category c.

[0134] In this embodiment, the matching results are used to distinguish the processing scenarios. The mapping table is used to ensure the accuracy of known candidate words, and the preset values ​​are used to cover the completeness of unknown candidate words. At the same time, the calculation logic is simplified to achieve high efficiency, and the adjustable parameter design achieves flexibility. Thus, the correlation between the candidate words of the first resource object and the category of the page resource object can be determined quickly and accurately.

[0135] In one embodiment, obtaining the page resource object category of the page browsed by the user object in step S201 may include the following steps:

[0136] Identify the page viewed by the user object and extract the page's category identification information; the category identification information includes at least one of page attribute information and page content; determine the page resource object category based on the category identification information.

[0137] In the specific implementation, after detecting the user object's input operation, the pages that the user object has browsed can be determined, and category identification information can be extracted. The category identification information is used to identify the category to which the resource objects displayed on the page belong.

[0138] In this embodiment, the page category identification information may include at least one of page attribute information and page content. The page attribute information may be inherent to the page itself and used to identify the page's basic characteristics, such as page data tags, URL paths, preset category identifiers, or one or more other types. The page content may be specific information presented to the user on the page, such as the resource object name in a resource object details page, or the article text and keywords in an information page.

[0139] Furthermore, the category of page resource objects can be determined based on the category identification information. For example, the extracted category identification information can be matched with a preset category classification system for various resource objects, and the category of page resource objects corresponding to each page can be determined through rule matching, keyword comparison, and other methods.

[0140] In some possible embodiments, after detecting user input, the AccessibilityService service can be used to listen to the UI hierarchy of the currently focused window and extract the following fields: (1) Page title; for example, it can be obtained through the window.getTitle() function; (2) Product category tag; such as an element in the DOM with data-category="perfume", or text such as "belonging to category: perfume" in the page recognized by OCR); (3) Meta tags, such as <meta property="product:category" content="美妆 / 香水"> This can be obtained through the WebView client interface. Then, one or more standardized product category identifiers can be output, mapped to a predefined category system (such as an e-commerce general category tree). It is important to emphasize that, considering the data security and privacy of user objects, sensitive data such as raw HTML, URLs, and screenshots should neither be stored nor analyzed and uploaded. For example, by extracting information from the above fields, the following extraction results can be obtained:

[0141] {

[0142] "context_type": "PAGE_CATEGORY",

[0143] "categories": [

[0144] {

[0145] "category_id": "100203",

[0146] "category_path": "Beauty > Fragrance > Men's Fragrance",

[0147] "source": "meta_tag" # Meta tags return the most granular category.

[0148] },

[0149] {

[0150] "category_id": "100200",

[0151] "category_path": "Beauty > Perfume",

[0152] "source": "dom_text" # The DOM tag displays the navigation path on the page.

[0153] }

[0154] ],

[0155] "timestamp": 1733045600, / / Unix timestamp (seconds)

[0156] Page title: AAA Brand Men's Eau de Parfum 100ml - XX International

[0157] }

[0158] Here, category_id represents the category to which the product described on the currently viewed page belongs.

[0159] In this embodiment, the category identification information may include at least one of page attribute information and page content, providing diverse identification criteria for identifying the category of page resource objects. When the page attribute information is incomplete, it can be supplemented by the page content, and vice versa, which improves the reliability of category identification and helps to accurately associate candidate words with the category of page resource objects required by the user, thereby improving the relevance of the candidate word association results to the user's current scene.

[0160] In one embodiment, in step S202, determining multiple first resource object candidate words containing the candidate word prefix, using the input information as a candidate word prefix, may include:

[0161] Obtain the prefix tree; the prefix tree includes multiple resource object candidate words; search the prefix tree using the input information as the prefix of the candidate words, and obtain multiple first resource object candidate words based on the search results.

[0162] In the specific implementation, a prefix tree can be obtained first. As a pre-constructed tree-like data structure, the prefix tree internally stores multiple candidate words for resource objects. These candidate words can be pre-organized and stored based on actual application scenario requirements, such as resource objects involved in user history interactions. For example, in the tree-like data structure of the prefix tree, each node represents a character, and all characters on the path from the root node to a certain node can be concatenated to form a string (such as a candidate word for a resource object). This structure can support efficient prefix matching queries.

[0163] After obtaining the prefix tree, input information can be used as candidate word prefixes to perform a search operation within the prefix tree. Specifically, the input information can be text snippets, keyword combinations, or other parts of the content indicating the desired resource object entered by the user in the interactive interface. Using this input information as a candidate word prefix for searching can filter out all resource object candidate words in the prefix tree that begin with that input information. Subsequently, based on the search results in the prefix tree, multiple first resource object candidate words that meet the requirements can be obtained. For example, a standard prefix tree can be used to store all product names. After the user enters a prefix (such as "fragrant"), the set of all first resource object candidate words {W1, W2, ..., Wn} that begin with the prefix can be quickly retrieved, such as perfume, melon, and aromatherapy.

[0164] In some optional embodiments, in order to further improve the accuracy and relevance of the first resource object candidate words, the searched resource object candidate words can be further filtered, for example, by sorting or filtering based on information such as the popularity of resource objects and user preference tags.

[0165] In this embodiment, by using the user's input information as a prefix and matching it in the prefix tree, multiple candidate words for the first resource object are obtained. On the one hand, the candidate word association can reduce the amount of information that the user needs to continue to input. On the other hand, when performing candidate word association in the future, the input content associated with the user's current application scenario can be effectively identified by combining the page resource object category, thereby improving the accuracy of the association of the remaining input content.

[0166] In one embodiment, obtaining the page resource object category of the page browsed by the user object in step S201 may include the following steps:

[0167] Determine the page resource object category of the first page currently viewed by the user object; and / or, determine the page resource object category of the second page previously viewed by the user object; the second page is a different page from the first page currently viewed by the user object, and the input operation is triggered on the first page.

[0168] In practical applications, when obtaining the page resource object category of the page browsed by a user object, this can be achieved by determining the page resource object category of the page currently being browsed by the user object. For ease of distinction, the page currently being browsed by the user object is referred to as the first page.

[0169] For example, when a user is browsing a specific page (i.e., the first page) on a terminal device, the system can analyze the first page in real time, identify and extract the category information corresponding to the page resource objects contained on the page, thereby obtaining the category of the page resource objects currently being browsed by the user.

[0170] By identifying the category of page resource objects on the first page currently being viewed by the user, the adaptability of candidate word association results to the user's current page interaction scenario can be effectively improved, thereby increasing the accuracy of candidate word association results.

[0171] Furthermore, this can be achieved by determining the page resource object categories of pages previously viewed by the user. For ease of distinction, pages previously viewed by the user but not currently viewed are referred to as "second pages." That is, the second page is different from the first page currently viewed by the user; the second page is a historical page that the user viewed before viewing the first page. Specifically, with the user's authorization, the user's browsing history can be retrieved, and one or more previously viewed second pages can be selected. These previously viewed second pages can be historical pages viewed within a preset time period. For example, the preset time period can be determined based on the user's search habits, such as the time range within which the user is expected to perform a resource object search after viewing the page. These second pages are then processed individually to identify the page resource object category corresponding to each second page, thereby achieving the acquisition of the resource object categories of pages previously viewed by the user.

[0172] By analyzing the second pages that users have previously visited and determining the categories of page resource objects on those pages, we can flexibly generate candidate word suggestions based on users' past page browsing behavior, reducing the input operations that users need to perform when entering information across pages or platforms (such as searching on the current page based on past page content).

[0173] Based on the identification of page resource object categories on the first or second page in this embodiment, the accuracy of candidate word association results can be effectively improved. Table 2 below provides several possible scenarios as examples:

[0174] Table 2

[0175]

[0176] In one embodiment, determining at least one second resource object candidate word from a plurality of first resource object candidate words based on relevance can include the following steps:

[0177] Obtain the search popularity of each first resource object candidate word; for each first resource object candidate word, determine the recommendation degree of the first resource object candidate word based on the relevance of the first resource object candidate word and the search popularity; determine at least one first resource object candidate word whose recommendation degree meets the preset recommendation degree condition as a second resource object candidate word.

[0178] In practice, the search popularity of each candidate term for the first resource object can be obtained. The search popularity can be determined based on the global search volume of the candidate term for the first resource object. Search popularity reflects the frequency with which the candidate term for the first resource object is viewed and retrieved by users in a global or general context. It can be obtained by relying on existing search engine data statistics systems, user behavior analysis platforms, or relevant resource retrieval databases, etc., by collecting and statistically analyzing search records and access data over a period of time, thereby obtaining the search popularity value corresponding to each candidate term for the first resource object.

[0179] In one embodiment, search popularity can be determined based on the global relevance of the first resource object candidate terms. The global relevance can be determined by querying a relevance table, which can be obtained in a similar way to the relevance mapping table. For example, historical search data of multiple sample user objects can be obtained. This historical search data can include resource object descriptors of resource objects selected by sample user objects through the platform homepage or website homepage (historical pages corresponding to non-page resource object categories). Search popularity can be determined by statistically analyzing co-occurrence frequency.

[0180] Subsequently, for each candidate term of the first resource object, the recommendation degree of the candidate term can be determined based on its existing relevance and the search popularity obtained in the above steps. In this process, relevance reflects the close connection between the candidate term and the user's application scenario, while search popularity reflects the degree of attention and potential demand for the candidate term within the overall global context. In some embodiments, a preset algorithm model can be used to comprehensively calculate the relevance and search popularity. For example, relevance and search popularity can be assigned certain weight coefficients and then weighted and summed, or other mathematical operations that can reasonably integrate the two information can be used to finally obtain the recommendation degree of each candidate term of the first resource object. This recommendation degree can intuitively reflect the priority and suitability of the candidate term in the recommendation process. For example, the comprehensive score Score(w) can be calculated as follows, which is the recommendation degree of the candidate term of the first resource object:

[0181] Score(w) =

[0182] Where global_pop(w) is the global popularity of the word, and α is the context weight coefficient. This weight can be adjusted according to the actual scenario. For example, the default value of α is 0.7, which can be dynamically increased to 0.9 in e-commerce applications.

[0183] Finally, candidate words for the first resource object that meet the preset recommendation criteria can be identified as candidate words for the second resource object. The preset recommendation criteria can be flexibly set according to actual application needs, technical scenarios, and recommendation goals. For example, this criterion could be a recommendation score greater than or equal to a preset threshold, or it could be selecting the top-ranked candidate words after sorting all first resource object candidate words in descending order of recommendation score. For instance, they could be sorted in descending order of Score(w), and the Top-K (e.g., K=6) first resource object candidate words could be selected as the final second resource object candidate words.

[0184] In some optional embodiments, the preset recommendation criteria can also be adjusted in conjunction with specific business rules. For example, different recommendation criteria can be set for different types of resource objects to ensure that the final determined second resource object candidate words can better meet the recommendation needs in actual applications and improve the accuracy and effectiveness of resource recommendations.

[0185] In this embodiment, by combining relevance and search popularity to calculate the recommendation degree, the relevance and popularity of candidate words are considered simultaneously, avoiding recommendation bias caused by single-dimensional screening, improving the overall quality of candidate words for the second resource object, and effectively improving the accuracy of candidate word association results.

[0186] To enable those skilled in the art to better understand the above steps, the embodiments of this application are illustrated below with some examples, but it should be understood that the embodiments of this application are not limited thereto.

[0187] In Example 1, such as Figure 5 As shown, this example may include the following steps:

[0188] S501, responding to input operations from user objects, determines the input information.

[0189] S502, determine the page browsed by the user object and extract the page's category identification information; determine the page resource object category based on the category identification information.

[0190] For example, the page resource object category of the first page currently being viewed by the user object can be determined, or the page resource object category of the second page previously viewed by the user object can be determined.

[0191] S503, using the input information as a prefix for candidate words, search the prefix tree and obtain multiple candidate words for the first resource object based on the search results.

[0192] S504, obtain the pre-built relevance mapping table, and determine the relevance between each first resource object candidate word and the page resource object category based on the matching result of each first resource object candidate word and the relevance mapping table.

[0193] The correlation mapping table can be constructed through the following steps:

[0194] Obtain historical search data for multiple sample user objects; based on the statistical results of the historical search data, determine the co-occurrence frequency between page resource object categories and each resource object descriptor, and determine the correlation between each resource object descriptor and page resource object categories based on the co-occurrence frequency; construct a correlation mapping table based on the correlation of multiple resource object descriptors.

[0195] S505, for each candidate word of the first resource object, determine the recommendation degree of the candidate word of the first resource object based on the relevance and search popularity of the candidate word of the first resource object; determine at least one candidate word of the first resource object that meets the preset recommendation degree condition as a candidate word of the second resource object.

[0196] S506, obtain the candidate word association results of the input information based on the candidate words of the second resource object.

[0197] In Example 2, suppose a user is browsing the product page "AAA Brand Men's Eau de Parfum 100ml" on the XX platform, categorized as "Beauty → Perfume → Men's Perfume." Five minutes later, the user types "fragrance" into the instant messaging chat box. In this example, when the user types "fragrance," the system prioritizes recommending product names highly relevant to men's perfumes, such as perfume, rather than information like sausage, which might be globally popular but irrelevant to the current context.

[0198] (1) Identify the category of page resource objects.

[0199] The following category identification information can be obtained:

[0200] {

[0201] "context_type": "PAGE_CATEGORY",

[0202] "categories": [

[0203] {

[0204] "category_id": "100203",

[0205] "category_path": "Beauty > Fragrance > Men's Fragrance",

[0206] "source": "meta_tag

[0207] }

[0208] ],

[0209] "timestamp": 1733045600,

[0210] Page title: Chanel Men's Eau de Parfum 100ml - XX International

[0211] }

[0212] Therefore, the product category is "100203".

[0213] (2) Obtain the correlation mapping table.

[0214] The association mapping table corresponding to page "100203" includes the following information:

[0215] {

[0216] "100203": {

[0217] "Perfume": 0.96,

[0218] Eau de Parfum: 0.94

[0219] "Eau de Cologne": 0.90,

[0220] Fragrance: 0.75

[0221] "Soap": 0.12,

[0222] "Sausage": 0.08

[0223] }

[0224] }

[0225] On the other hand, the global popularity table global_pop(w) is obtained. The information in the table is determined based on the search frequency across the entire platform, as shown in Table 3 below:

[0226] Table 3

[0227]

[0228] (3) Determine multiple candidate words for the first resource object based on the input information.

[0229] Since the user inputs "fragrance", the system retrieves all product names starting with "fragrance" from the local prefix tree. If there are 6, Candidates = {perfume, soap, sausage, fragrance, perfume, aromatherapy}.

[0230] (4) Determine the recommendation level and dynamically re-rank.

[0231] Assuming the context weight coefficient α = 0.7, the calculated recommendation score for some candidate words is shown in Table 4 below:

[0232] Table 4

[0233]

[0234] Finally, the candidate words are sorted from highest to lowest score and output as a whole.

[0235] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0236] Based on the same inventive concept, this application also provides a candidate word association device for implementing the candidate word association method described above. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more candidate word association device embodiments provided below can be found in the limitations of the candidate word association method above, and will not be repeated here.

[0237] In one exemplary embodiment, such as Figure 6 As shown, a candidate word association device is provided, comprising:

[0238] The category recognition module 601 is used to respond to the input operation of the user object, determine the input information, and obtain the page resource object category of the page browsed by the user object; the page resource object category is determined according to the category to which the resource object displayed on the page belongs;

[0239] The correlation determination module 602 is used to determine a plurality of first resource object candidate words containing the candidate word prefix by using the input information as the candidate word prefix, and to determine the correlation between each first resource object candidate word and the page resource object category;

[0240] The candidate word filtering module 603 is used to determine at least one second resource object candidate word from the plurality of first resource object candidate words based on the relevance.

[0241] The association result determination module 604 is used to obtain and display the candidate word association results of the input information based on the candidate words of the second resource object.

[0242] In one embodiment, the correlation determination module 602 is configured to:

[0243] Obtain a pre-built association mapping table; the association mapping table records the association degree between multiple resource object descriptors and the page resource object categories;

[0244] Based on the matching results between each candidate word of the first resource object and the relevance mapping table, the relevance between each candidate word of the first resource object and the category of the page resource object is determined.

[0245] In one embodiment, the correlation determination module 602 is further configured to:

[0246] Obtain historical search data for each of multiple sample user objects; the historical search data includes historical pages corresponding to the page resource object categories in which the sample user objects have performed input operations, and resource object descriptors corresponding to the resource objects selected by the sample user objects based on the historical pages;

[0247] Based on the statistical results of the historical search data, the co-occurrence frequency of the page resource object category and each resource object descriptor is determined, and the correlation between each resource object descriptor and the page resource object category is determined based on the co-occurrence frequency.

[0248] A correlation mapping table is constructed based on the correlation degree of each of the resource object descriptors.

[0249] In one embodiment, the correlation determination module 602 is configured to:

[0250] If the correlation mapping table contains resource object descriptive words that match the first resource object candidate words, then the correlation between the first resource object candidate words and the page resource object category is determined based on the correlation of the matching resource object descriptive words in the correlation mapping table.

[0251] If there is no resource object description word matching the first resource object candidate word in the correlation mapping table, the correlation between the first resource object candidate word and the page resource object category is determined according to a preset value less than the correlation threshold.

[0252] In one embodiment, the category identification module 601 is used for:

[0253] The page browsed by the user object is determined, and the category identification information of the page is extracted; the category identification information includes at least one of page attribute information and page content;

[0254] The page resource object category of the page is determined based on the category identification information.

[0255] In one embodiment, the correlation determination module 602 is configured to:

[0256] Obtain the prefix tree; the prefix tree includes multiple resource object candidate words;

[0257] Using the input information as a prefix for candidate words, a search is performed in the prefix tree, and multiple candidate words for first resource objects are obtained based on the search results.

[0258] In one embodiment, the category identification module 601 is used for:

[0259] Determine the page resource object category of the first page currently being viewed by the user object; and / or,

[0260] Determine the page resource object category of the second page previously viewed by the user object; the second page is different from the first page currently viewed by the user object, and the input operation is triggered on the first page.

[0261] In one embodiment, the candidate word filtering module 603 is used for:

[0262] Obtain the search popularity of each candidate word for the first resource object;

[0263] For each candidate word of the first resource object, the recommendation degree of the candidate word of the first resource object is determined based on the relevance and search popularity of the candidate word of the first resource object.

[0264] At least one of the first resource object candidate words whose recommendation level meets the preset recommendation level conditions is determined as the second resource object candidate word.

[0265] Each module in the aforementioned candidate word association device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0266] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 7 As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When executed by the processor, the computer program implements a candidate word association method. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0267] Those skilled in the art will understand that Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0268] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0269] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0270] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0271] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0272] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0273] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0274] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A candidate word association method, characterized in that, The method includes: In response to an input operation by a user object, the input information is determined, and the page resource object category of the page browsed by the user object is obtained; the page resource object category is determined according to the category to which the resource objects displayed on the page belong. Using the input information as a candidate word prefix, determine multiple first resource object candidate words containing the candidate word prefix, and determine the correlation between each first resource object candidate word and the page resource object category; Based on the relevance, at least one second resource object candidate word is determined from the plurality of first resource object candidate words; The candidate word association results of the input information are obtained and displayed based on the candidate words of the second resource object.

2. The method according to claim 1, characterized in that, Determining the correlation between each candidate word of the first resource object and the category of the page resource object includes: Obtain a pre-built association mapping table; the association mapping table records the association degree between multiple resource object descriptors and the page resource object categories; Based on the matching results between each candidate word of the first resource object and the relevance mapping table, the relevance between each candidate word of the first resource object and the category of the page resource object is determined.

3. The method according to claim 2, characterized in that, The correlation mapping table is constructed through the following steps: Obtain historical search data for each of multiple sample user objects; the historical search data includes historical pages corresponding to the page resource object categories in which the sample user objects have performed input operations, and resource object descriptors corresponding to the resource objects selected by the sample user objects based on the historical pages; Based on the statistical results of the historical search data, the co-occurrence frequency of the page resource object category and each resource object descriptor is determined, and the correlation between each resource object descriptor and the page resource object category is determined based on the co-occurrence frequency. A correlation mapping table is constructed based on the correlation degree of each of the resource object descriptors.

4. The method according to claim 2, characterized in that, The step of determining the relevance between each candidate word of the first resource object and the category of the page resource object based on the matching result of each candidate word of the first resource object and the relevance mapping table includes: If the correlation mapping table contains resource object descriptive words that match the first resource object candidate words, then the correlation between the first resource object candidate words and the page resource object category is determined based on the correlation of the matching resource object descriptive words in the correlation mapping table. If there is no resource object description word matching the first resource object candidate word in the correlation mapping table, the correlation between the first resource object candidate word and the page resource object category is determined according to a preset value less than the correlation threshold.

5. The method according to claim 1, characterized in that, The process of obtaining the page resource object category of the page browsed by the user object includes: The page browsed by the user object is determined, and the category identification information of the page is extracted; the category identification information includes at least one of page attribute information and page content; The page resource object category of the page is determined based on the category identification information.

6. The method according to claim 1, characterized in that, The step of determining multiple first resource object candidate words containing the candidate word prefix using the input information as a candidate word prefix includes: Obtain the prefix tree; the prefix tree includes multiple resource object candidate words; Using the input information as a prefix for candidate words, a search is performed in the prefix tree, and multiple candidate words for first resource objects are obtained based on the search results.

7. The method according to claim 1, characterized in that, The process of obtaining the page resource object category of the page browsed by the user object includes: Determine the page resource object category of the first page currently being viewed by the user object; and / or, Determine the page resource object category of the second page previously viewed by the user object; the second page is different from the first page currently viewed by the user object, and the input operation is triggered on the first page.

8. The method according to any one of claims 1 to 7, characterized in that, The step of determining at least one second resource object candidate word from the plurality of first resource object candidate words based on the relevance includes: Obtain the search popularity of each candidate word for the first resource object; For each candidate word of the first resource object, the recommendation degree of the candidate word of the first resource object is determined based on the relevance and search popularity of the candidate word of the first resource object. At least one of the first resource object candidate words whose recommendation level meets the preset recommendation level conditions is determined as the second resource object candidate word.

9. A candidate word association device, characterized in that, The device includes: The category recognition module is used to respond to the input operation of the user object, determine the input information, and obtain the page resource object category of the page browsed by the user object; the page resource object category is determined according to the category to which the resource object displayed on the page belongs; The relevance determination module is used to determine multiple first resource object candidate words containing the candidate word prefix by using the input information as a candidate word prefix, and to determine the relevance between each first resource object candidate word and the page resource object category; The candidate word filtering module is used to determine at least one second resource object candidate word from the plurality of first resource object candidate words based on the relevance. The association result determination module is used to obtain and display the candidate word association results of the input information based on the candidate words of the second resource object.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 8.

Images