Information recommendation methods, devices and electronic equipment

Abstract

This disclosure provides an information recommendation method, apparatus, and electronic device, relating to the field of data processing technology, specifically to the field of search recommendation technology. The specific implementation scheme is as follows: N historical search messages matching the user-input target search information are obtained from log data; a list of synonyms for words in the N historical search messages is determined; if the N historical search messages include a first historical search message, the first historical search message is deleted from the N historical search messages to obtain first candidate recommendation information. The first historical search message includes a first word, and the list of synonyms for the first word overlaps with the list of synonyms for a second word. The second word includes at least one of the following: a word from the first historical search message, a word from the second historical search message in the N historical search messages corresponding to the first word, and a word from the target search message corresponding to the first word; information recommendation is performed based on the first candidate recommendation information.

Description

Technical Field

[0001] This disclosure relates to the field of data processing technology, and more particularly to the field of search and recommendation technology, specifically to an information recommendation method, apparatus, and electronic device. Background Technology

[0002] With the development of science and technology and Internet technology, e-commerce platforms have risen rapidly. E-commerce platforms can provide users with a variety of online services, bringing great convenience to production and life.

[0003] On e-commerce platforms, when users enter a site to search for products, they usually need to enter search information in the search bar. Accordingly, the platform can make information recommendations based on the user's search information.

[0004] Currently, information recommendation methods typically involve extracting historical search information related to the user's input search information from log data to make recommendations. Summary of the Invention

[0005] This disclosure provides an information recommendation method, apparatus, and electronic device.

[0006] According to a first aspect of this disclosure, an information recommendation method is provided, comprising:

[0007] Retrieve N historical search results from the log data that match the user's target search information, where N is a positive integer;

[0008] Determine a list of synonyms for the words in the N historical search results;

[0009] If the first historical search information is included in the N historical search information, the first historical search information is deleted from the N historical search information to obtain the first candidate recommendation information. The first historical search information includes a first word, and the synonym list of the first word and the synonym list of the second word have an intersection. The second word includes at least one of the following: the word in the first historical search information, the word in the second historical search information in the N historical search information that corresponds to the first word, and the word in the target search information that corresponds to the first word.

[0010] Information recommendations are made based on the first candidate recommendation information.

[0011] According to a second aspect of this disclosure, an information recommendation device is provided, comprising:

[0012] The first acquisition module is used to obtain N historical search information that match the target search information entered by the user from the log data, where N is a positive integer;

[0013] The first determining module is used to determine a list of synonyms for words in the N historical search information;

[0014] The deletion module is used to delete the first historical search information from the N historical search information when the first historical search information is included in the N historical search information, to obtain the first candidate recommendation information. The first historical search information includes a first word, and the synonym list of the first word and the synonym list of the second word have an intersection. The second word includes at least one of the following: the word in the first historical search information, the word in the second historical search information corresponding to the first word in the N historical search information, and the word in the target search information corresponding to the first word.

[0015] The recommendation module is used to recommend information based on the first candidate recommendation information.

[0016] According to a third aspect of this disclosure, an electronic device is provided, comprising:

[0017] At least one processor; and

[0018] A memory that is communicatively connected to at least one processor; wherein,

[0019] The memory stores instructions that can be executed by at least one processor to enable the at least one processor to perform any of the methods in the first aspect.

[0020] According to a fourth aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions for causing a computer to perform any of the methods in the first aspect.

[0021] According to a fifth aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements any of the methods in the first aspect.

[0022] The technology disclosed herein solves the problem of poor information recommendation performance and improves the effectiveness of information recommendation.

[0023] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0024] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0025] Figure 1 This is a flowchart illustrating the information recommendation method according to the first embodiment of this disclosure;

[0026] Figure 2 This is a schematic diagram of the structure of an information recommendation device according to a second embodiment of the present disclosure;

[0027] Figure 3 This is a schematic block diagram of an example electronic device used to implement embodiments of the present disclosure. Detailed Implementation

[0028] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0029] First Embodiment

[0030] like Figure 1 As shown, this disclosure provides an information recommendation method, including the following steps:

[0031] Step S101: Obtain N historical search results from the log data that match the target search information entered by the user.

[0032] Where N is a positive integer.

[0033] In this embodiment, the information recommendation method relates to the field of data processing technology, and particularly to the field of search recommendation technology, which can be widely applied in product procurement scenarios. The information recommendation method of this embodiment can be executed by the information recommendation device of this embodiment. The information recommendation device of this embodiment can be configured in any electronic device to execute the information recommendation method of this embodiment. The electronic device can be a server or a terminal device; no specific limitation is made here.

[0034] This embodiment can be applied to both B2B and B2C procurement scenarios, without specific limitations. The difference between B2B and B2C procurement scenarios lies in the different users targeted by the e-commerce platforms. B2B procurement typically targets groups, such as communities, organizations, or companies, while B2C procurement typically targets individuals.

[0035] The target search information can be the information that the user enters in the search input box, such as the user entering the target search information "notebook" on an e-commerce platform.

[0036] When the information recommendation device detects that a user is typing in the search input box, it can acquire the user's target search information in real time or periodically, and recommend information based on this target search information. The device can predict the search information the user might need based on this target search information and display the predicted search information in a drop-down menu in the search input box for information recommendation. This provides users with a more precise expression of their needs, helps them quickly locate the content they want to search for, and saves them typing time.

[0037] When the user inputs the target search information, the information recommendation device can match the target search information with the historical search information in the log data to obtain N historical search information that match the target search information from the log data, where N is a positive integer.

[0038] Before matching, historical search information in the log data can be filtered, or after matching, multiple historical search information that match the target search information can be filtered to perform quality control on the search information before information recommendation, resulting in N historical search information.

[0039] Quality control may include: 1) using a pre-established blacklist of search information to filter historical search information, that is, filtering out historical search information in the blacklist and filtering out historical search information containing words in the blacklist; 2) filtering out historical search information that is too short or too long; 3) filtering out historical search information that has too few search results or is not highly relevant to the product.

[0040] Matching historical search information with target search information can refer to semantic matching, matching of categories corresponding to historical search information with categories corresponding to target search information, or matching of core words in historical search information with core words in target search information. No specific limitations are made here.

[0041] Accordingly, the methods for matching the target search information with historical search information in the log data include, but are not limited to, semantic matching, category matching, and core word matching.

[0042] Step S102: Determine a list of synonyms for the words in the N historical search information.

[0043] In this step, each of the N historical search results includes at least one word. Historical search results can include nouns, adjectives, or verbs. For example, if the historical search result is "small notebook", then the historical search result can include adjectives and nouns. Or, if the historical search result is "folding airplane", then the historical search result can include verbs and nouns.

[0044] A list of synonyms for words in N historical search information can be determined. In an optional implementation, for each historical search information, word segmentation and query processing can be performed on the historical search information based on a pre-trained model to obtain the target result of the historical search information. The target result may include at least one word of the historical search information and a list of synonyms for each word.

[0045] A list of synonyms for a word can include words that are synonymous with that word. For example, a list of synonyms for the word "custom-made" can include words such as "customization" and "production".

[0046] Step S103: If the first historical search information is included in the N historical search information, delete the first historical search information from the N historical search information to obtain the first candidate recommendation information. The first historical search information includes a first word. The synonym list of the first word and the synonym list of the second word have an intersection. The second word includes at least one of the following: the word in the first historical search information, the word in the second historical search information in the N historical search information that corresponds to the first word, and the word in the target search information that corresponds to the first word.

[0047] The first historical search information can be historical search information with semantic repetition within the information. For example, the historical search information "notebook / notepad small" contains the two words "notebook" and "notepad" which are repeated.

[0048] The first historical search information can also be historical search information that has semantic overlap with other historical search information in the N historical search information. For example, the first historical search information "customized notebook" has semantic overlap with other historical search information "customized notebook".

[0049] The first historical search information can also be historical search information that has semantic overlap with the target search information.

[0050] The existence of a semantically duplicated first historical search result among N historical search results can be determined by cross-validating the synonym lists of two words. Specifically, when identifying historical search results with internal semantic duplication, the synonym lists of two words within the historical search results can be cross-validated. If the synonym lists of two words overlap, then the historical search result contains internal semantic duplication and is considered the first historical search result. In this case, both the first and second words are two words from the same historical search result.

[0051] For example, in the historical search information "notebook, small notebook", there is an overlap between the list of synonyms for the first word "notebook" and the list of synonyms for the second word "notebook", which constitutes the first historical search information.

[0052] When identifying historical search information with semantic overlap, the synonym lists of corresponding words in two historical search pieces can be cross-validated. If the synonym lists of every two corresponding words overlap, then the two historical search pieces have semantic overlap, and one of them is designated as the first historical search piece. In this case, the first word and the second word are the two corresponding words in the two historical search pieces, respectively.

[0053] For example, if one historical search result is "custom-made notebooks" and another is "customized notebooks", there is an overlap between the list of synonyms for the first word "custom-made" and the list of synonyms for the second word "customized". These two historical search results contain semantic overlap.

[0054] Furthermore, determining whether there is semantic overlap between the target search information and historical search information can be done in the same way as determining historical search information with semantic overlap, and will not be elaborated here. In this case, the first word is a word in the historical search information, and the second word can be a word in the target search information that corresponds to the first word.

[0055] Among these, the correspondence between two words in different search results may include, but is not limited to, location correspondence, part-of-speech correspondence, etc.

[0056] If the first historical search information is included in the N historical search results, the first historical search information can be deleted from the N historical search results to obtain the first candidate recommendation information. The first candidate recommendation information can include the historical search information after deleting the first historical search information from the N historical search results. The historical search information in the first candidate recommendation information can be used as the search information to be recommended for information recommendation.

[0057] Step S104: Recommend information based on the first candidate recommendation information.

[0058] In this step, the first candidate recommendation information may include one, two or more search information to be recommended. The recommendation weight of each search information to be recommended can be determined. In an optional implementation, the recommendation weight of the search information to be recommended can be determined based on the degree of matching between the search information to be recommended and the target search information. The search information to be recommended is sorted in descending order of recommendation weight, and the search information to be recommended with the highest recommendation weight is recommended to the user.

[0059] In this embodiment, when determining the search information to be recommended based on historical search information, the existence of semantically duplicated historical search information can be determined by cross-validating the lists of synonyms of the two words, and the semantically duplicated historical search information can be deleted, thereby optimizing the search information to be recommended and improving the information recommendation effect.

[0060] Optionally, prior to step S103, the method further includes at least one of the following:

[0061] For each of the N historical search messages, if the synonym lists of any two words in the historical search messages have an intersection, then the historical search message is determined to be the first historical search message.

[0062] If a first target historical search information exists among the N historical search information, the first target historical search information is determined to be the first historical search information. For each word in the first target historical search information, there is a word corresponding to the word in the target search information, and the synonym lists of every two corresponding words in the first target historical search information and the target search information have an intersection.

[0063] When N is greater than 1, for every two historical search messages in the N historical search messages, if the synonym lists of every two corresponding words in the two historical search messages have an intersection, then one of the historical search messages in the two historical search messages is determined as the first historical search message, and the two corresponding words in the two historical search messages come from the two historical search messages respectively.

[0064] In this embodiment, semantic repetition can be detected within each of the N historical search messages. If the historical search message includes at least two words, the synonym lists of each pair of words in the historical search message can be cross-validated. If the synonym lists of any two words have an intersection, then the historical search message has semantic repetition within the message and is designated as the first historical search message.

[0065] It can detect semantic repetition between target search information and each of N historical search information. When there is semantic repetition between historical search information and target search information, the number of words in the target search information is usually greater than or equal to the number of words in the historical search information.

[0066] Therefore, for each word in the historical search information, it can be determined whether there is a corresponding word in the target search information. If there is, it can be determined whether there is an overlap in the synonym lists of each pair of corresponding words. If there is an overlap, it can be determined that there is semantic duplication between the historical search information and the target search information.

[0067] When N is greater than 1, semantic duplication can be detected between every two historical search messages in the N historical search messages. The synonym lists of the two corresponding words in the two historical search messages can be cross-validated. If the synonym lists of every two corresponding words have an intersection, then there is semantic duplication between the two historical search messages, and one of the two historical search messages is the first historical search message.

[0068] When detecting semantic redundancy between information, if only the inclusion relationship of synonym lists of any two corresponding words is used for deduplication, multiple search results containing the core word may be identified as synonyms. For example, the target search result "notebook", a previous search result "custom notebook", and another previous search result "notebook customization" may all be identified as synonyms because they all contain the core word "notebook".

[0069] Therefore, in practical applications, words in historical search results that contain synonyms of the core words of the target search information can be removed. The remaining words can then be used for cross-validation of synonym lists. If the synonym lists of every two corresponding words overlap, then the two search results are considered to have semantic overlap. This reduces computational complexity.

[0070] For example, if the target search information is "notebook", and one historical search information is "custom-made notebook" and another historical search information is "customized notebook", the word "notebook" can be removed from these two historical search information. The synonym lists of the remaining two corresponding words "custom-made" and "customized" can be cross-validated. If the synonym lists of these two words have an intersection, it is determined that there is semantic duplication between these two historical search information.

[0071] In this embodiment, by cross-validating the lists of synonyms for two words, it is possible to detect semantic repetition within the first historical search information and semantic repetition between information in N historical search information.

[0072] Optionally, before step S103, the method further includes:

[0073] If a second target historical search information exists among the N historical search information, the second target historical search information is determined to be the first historical search information, and the target search information includes the second target historical search information.

[0074] In this embodiment, for each of the N historical search information, it can be determined whether the target search information and the historical search information have an information inclusion relationship. If the target search information includes the historical search information, the historical search information can be determined as the first historical search information.

[0075] The inclusion of historical search information in the target search information means that the target search information contains the same content as historical search information. For example, if the target search information is "hardcover notebook" and a historical search information is "notebook", then the target search information includes that historical search information.

[0076] The target search information includes historical search information, or it can refer to the existence of semantically similar content in the target search information. For example, if the target search information is "large desktop computer" and a historical search information is "computer", then the target search information includes that historical search information.

[0077] In this embodiment, by detecting the information inclusion relationship between the target search information and the historical search information, it is possible to detect the first historical search information among N historical search information that has semantic repetition with the target search information.

[0078] Optionally, the target search information includes M words, and before step S104, the method further includes:

[0079] Identify the first core word among the M words;

[0080] Obtain the descriptive words associated with the first core word in the database, wherein the database stores the core word and the descriptive words;

[0081] The first core word is concatenated with the descriptive words associated with the first core word to obtain the second candidate recommendation information;

[0082] The information recommendation based on the first candidate recommendation information includes:

[0083] Information recommendations are made based on the first candidate recommendation information and the second candidate recommendation information.

[0084] In this embodiment, the target search information may include M words, where M is a positive integer. The M words may include nouns, adjectives, or verbs, etc. For example, if the target search information is "small notebook", then the target search information may include adjectives and nouns. Or, for example, if the target search information is "folding airplane", then the target search information may include verbs and nouns.

[0085] In one optional implementation, if the user enters target search information and separates different words with spaces, the information recommendation device can divide the words by detecting spaces to obtain M words.

[0086] In another alternative implementation, the information recommendation device can segment the target search information using a pre-trained word segmentation model, such as the jieba word segmentation tool, to obtain M words and the segmentation weight of each word. Before word segmentation, useless characters in the target search information can be filtered out using words from a pre-stored stop word list, and then word segmentation can be performed, which can improve the accuracy of word segmentation.

[0087] The first core word among M words can refer to the keyword among the M words. This keyword can be a noun, referring to the word that best expresses the user's search needs in the target search information. For example, if the target search information is "hard-shell notebook", then the keyword for the target search information is notebook.

[0088] There are several ways to determine the first core word. For example, you can determine whether the words in M ​​words match the words in the keyword database, and then determine the words in the M words that match the keyword database as the first core word.

[0089] For example, for each of the M words, category analysis can be performed to obtain the category information corresponding to the word; at least one candidate word can be determined from the M words, and the category information corresponding to each candidate word has an intersection with the category information obtained by category analysis of the target search information; the first core word can be determined from the at least one candidate word.

[0090] In this embodiment, the database stores core words and descriptive words in association, and a core word can be associated with one, two or more descriptive words. The descriptive words associated with the first core word in the database can be obtained, and the first core word can be concatenated with the descriptive words associated with the first core word to obtain the second candidate recommendation information.

[0091] For example, if the first core keyword is "notebook", the descriptive words associated with the first core keyword in the database include hard-shell, small, and extra-thick. By concatenating the first core keyword with each descriptive word, we can obtain the second candidate recommendation information, including the concatenated search information "hard-shell notebook", "small notebook", and "extra-thick notebook".

[0092] Before recommending information, the core words in the historical search information in the log data can be determined in advance. The words in the historical search information other than the core words are identified as descriptive words, and the core words and descriptive words are associated and stored in the database. This will be explained in more detail below.

[0093] After obtaining the second candidate recommendation information, information recommendations can be made based on the first and second candidate recommendation information. In an optional implementation, the first and second candidate recommendation information can be aggregated. After aggregation, since the first and second candidate recommendation information may overlap, i.e., there may be duplicate search information to be recommended, in this case, deduplication can be performed, and only one of the duplicate search information to be recommended can be kept.

[0094] In this implementation, the recommendation weight of each search result to be recommended can be determined, and the search results to be recommended can be sorted from largest to smallest according to the recommendation weight. The search results to be recommended with the highest recommendation weight are then recommended to the user.

[0095] If the search information to be recommended is among the first candidate search information, the recommendation weight of the search information to be recommended can be determined based on the degree of matching between the search information to be recommended and the target search information.

[0096] If the search information to be recommended is among the second-ranked candidate search information, its recommendation weight can be determined based on the weights corresponding to the descriptive terms in that search information. For example, the weights corresponding to the descriptive terms in the search information can be determined as the recommendation weight of that search information. In the database, each descriptive term associated with a core term can correspond to a weight; the higher the weight, the stronger the association between the descriptive term and the core term.

[0097] If the search information to be recommended is the search information to be recommended in the intersection of the first candidate recommendation information and the second candidate recommendation information, in this case, it can be determined based on the weight corresponding to the search information to be recommended and the weight corresponding to the descriptive words in the search information to be recommended. For example, the average of the weight corresponding to the search information to be recommended and the weight corresponding to the descriptive words in the search information to be recommended can be determined as the recommendation weight of the search information to be recommended.

[0098] In another optional implementation, a first recommendation weight can be determined based on a preset first channel weight for historical search information in the first candidate recommendation information; a second recommendation weight can be determined based on a preset second channel weight for information obtained by concatenating the first core word with the descriptive words associated with the first core word in the second candidate recommendation information; target recommendation information can be determined from the first and second candidate recommendation information based on the first and second recommendation weights; and information recommendation can be performed based on the target recommendation information. In this way, search-guided recommendations can be performed by combining log data and core words as two separate channels, thereby further improving the effectiveness of information recommendation.

[0099] In this implementation, the first channel weight can be a channel for information recommendation based on log data, and the second channel weight can be a channel for information recommendation based on core words. Both the first channel weight and the second channel weight can be preset. For example, the first channel weight can be set to 0.7 and the second channel weight can be set to 0.3.

[0100] Accordingly, the first recommendation weight can be determined by combining the weight of the first channel and the weight corresponding to the historical search information in the first candidate recommendation information, i.e., the search information to be recommended. The second recommendation weight can be determined by combining the weight of the second channel and the weight corresponding to the descriptive terms in the search information to be recommended in the second candidate recommendation information. For example, the two weights can be multiplied to obtain the recommendation weight of the search information to be recommended.

[0101] In one possible implementation, the search information to be recommended in the first candidate recommendation information can be sorted from largest to smallest according to a first recommendation weight, and the search information to be recommended that ranks higher according to the first recommendation weight can be determined as the target recommendation information. Alternatively, the search information to be recommended in the second candidate recommendation information can be sorted from largest to smallest according to a second recommendation weight, and the search information to be recommended that ranks higher according to the second recommendation weight can be determined as the target recommendation information.

[0102] When recommending information based on a defined target, if duplicate search information exists within the defined target recommendation information, deduplication can be performed before recommending the target recommendation information to the user.

[0103] In this embodiment, by combining the first candidate recommendation information and the second candidate recommendation information for information recommendation, the recommendation information can be effectively expanded and enriched, thereby further improving the effectiveness of information recommendation.

[0104] Optionally, M is greater than 1, and determining the first core word among the M words includes:

[0105] For each of the M words, perform category parsing on the word to obtain the category information corresponding to the word;

[0106] At least one candidate word is determined from the M words, and the category information corresponding to each candidate word has an intersection with the category information obtained by parsing the target search information.

[0107] The first core word is determined from the at least one candidate word.

[0108] In this embodiment, a pre-trained category parsing model can be used to parse the target search information into categories, thereby obtaining the category information corresponding to the target search information. Furthermore, for each of the M words, the same category parsing model can be used to parse the category of that word, obtaining the corresponding category information.

[0109] Among them, the category parsing model can use the third-level category as the parsing target, that is, the category information obtained by parsing is the third-level category, and the parsed category information can include at least one category and the weight corresponding to each category.

[0110] The category information corresponding to a word can include at least one category and the weight of each category. At least one candidate word can be determined from M words. The determination method is as follows: for each word in the M words, it can be determined whether the category information corresponding to the word intersects with the category information corresponding to the target search information. If there is an intersection, the word can be determined as a candidate word.

[0111] Then, the first core word can be determined from at least one candidate word. Specifically, the target weight of each candidate word can be determined. In an optional implementation, the word segmentation weight and position weight of the candidate word can be multiplied to obtain the target weight of the candidate word. The position weight of the candidate word can be assigned according to the position of the candidate word in the target search information; the earlier the position, the greater the position weight. Accordingly, the candidate word with the largest target weight among at least one candidate word can be determined as the first core word.

[0112] In this embodiment, the first core word in the target search information is determined by combining the category method. This can improve the accuracy of core word determination and thus further improve the information recommendation effect.

[0113] It should be noted that the core keywords in the database can also be determined using the same method described above. Specifically, log data can be obtained, which may include historical search information. A pre-established blacklist of search information can be used to filter historical search information in the log data, as can a pre-established category blacklist. For example, historical search information corresponding to categories that overlap with the category blacklist can be filtered out. Additionally, useless characters in historical search information can be filtered out using a pre-stored stop word list.

[0114] For the filtered log data, a word segmentation model can be used to segment the historical search information in the log data, obtaining segmentation results and word weights. If there is only one segmentation result, it is determined as the core word. If there are multiple segmentation results, the category information of each segmentation result is cross-validated with the category information of the historical search information to obtain candidate words. If there is only one candidate word, it is the core word. If there are multiple candidate words, they are weighted according to their position in the historical search information. The target weight of a candidate word = segmentation weight * position ratio. The candidate word with the highest target weight is taken as the core word, and the remaining segmentation results in the historical search information are used as descriptive words.

[0115] Subsequently, if the core words of different historical search messages in the log data are the same, the descriptive words associated with those core words can be clustered. This will result in a core word being associated with multiple descriptive words, and the core words and descriptive words can be stored in the database. Furthermore, the weight of each descriptive word can be determined based on its segmentation weight and positional proportion, and this weight can also be stored in the database.

[0116] In addition, the information recommendation device can also combine third-party candidate recommendation information for information recommendation. This third-party candidate recommendation information can include the user's historical search information, reflecting the user's personalized preferences. Based on the user's current and past historical search data, the device can mine the user's personalized preferences. Specifically, the user's current historical search data best expresses the user's current possible search preferences. Therefore, if the target search information is included in the current historical search data, the most recent historical search information is selected as the third-party candidate recommendation information in chronological order and recommended to the user first. If the target search information is included in past historical search data, it is sorted according to the historical click count of the target search information and added to the third-party candidate recommendation information.

[0117] Second Embodiment

[0118] like Figure 2As shown, this disclosure provides an information recommendation device 200, including:

[0119] The first acquisition module 201 is used to acquire N historical search information that match the target search information input by the user from the log data, where N is a positive integer;

[0120] The first determining module 202 is used to determine a list of synonyms for words in the N historical search information;

[0121] The deletion module 203 is used to delete the first historical search information from the N historical search information when the N historical search information includes the first historical search information, to obtain the first candidate recommendation information. The first historical search information includes a first word, and the synonym list of the first word and the synonym list of the second word have an intersection. The second word includes at least one of the following: the word in the first historical search information, the word in the second historical search information in the N historical search information that corresponds to the first word, and the word in the target search information that corresponds to the first word.

[0122] The recommendation module 204 is used to recommend information based on the first candidate recommendation information.

[0123] Optionally, the device further includes:

[0124] The second determining module is used to determine the historical search information as the first historical search information if, for each of the N historical search information, there is an intersection between the synonym lists of any two words in the historical search information.

[0125] The third determining module is used to determine the first target historical search information as the first historical search information when the first target historical search information exists in the N historical search information. For each word in the first target historical search information, there is a word corresponding to the word in the target search information, and the synonym lists of every two corresponding words in the first target historical search information and the target search information have an intersection.

[0126] The fourth determining module is used to determine, when N is greater than 1, for every two historical search messages in the N historical search messages, if the synonym lists of every two corresponding words in the two historical search messages have an intersection, one of the historical search messages in the two historical search messages is the first historical search message, and the two corresponding words in the two historical search messages are respectively from the two historical search messages.

[0127] Optionally, the device further includes:

[0128] The fifth determining module is used to determine the second target historical search information as the first historical search information when the second target historical search information exists among the N historical search information, wherein the target search information includes the second target historical search information.

[0129] Optionally, the target search information includes M words, and the device further includes:

[0130] The sixth determining module is used to determine the first core word among the M words;

[0131] The second acquisition module is used to acquire the descriptive words associated with the first core word in the database, wherein the database stores the core word and the descriptive words.

[0132] The splicing module is used to splice the first core word with the descriptive words associated with the first core word to obtain the second candidate recommendation information;

[0133] The recommendation module 204 is specifically used to recommend information based on the first candidate recommendation information and the second candidate recommendation information.

[0134] Optionally, if M is greater than 1, the sixth determining module is specifically used for:

[0135] For each of the M words, perform category parsing on the word to obtain the category information corresponding to the word;

[0136] At least one candidate word is determined from the M words, and the category information corresponding to each candidate word has an intersection with the category information obtained by parsing the target search information.

[0137] The first core word is determined from the at least one candidate word.

[0138] Optionally, the recommendation module 204 is specifically used for:

[0139] Based on the preset first channel weight, the first recommendation weight of the historical search information in the first candidate recommendation information is determined;

[0140] Based on the preset second channel weight, the second recommendation weight of the information obtained by splicing the first core word and the descriptive words associated with the first core word in the second candidate recommendation information is determined;

[0141] Based on the first recommendation weight and the second recommendation weight, target recommendation information is determined from the first candidate recommendation information and the second candidate recommendation information;

[0142] Information recommendations are made based on the target recommendation information.

[0143] The information recommendation device 200 provided in this disclosure can implement all the processes implemented in the information recommendation method embodiments and can achieve the same beneficial effects. To avoid repetition, it will not be described again here.

[0144] The collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0145] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0146] Figure 3 A schematic block diagram of an example electronic device that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0147] like Figure 3 As shown, device 300 includes a computing unit 301, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 302 or a computer program loaded from storage unit 308 into random access memory (RAM) 303. The RAM 303 may also store various programs and data required for the operation of device 300. The computing unit 301, ROM 302, and RAM 303 are interconnected via bus 304. Input / output (I / O) interface 305 is also connected to bus 304.

[0148] Multiple components in device 300 are connected to I / O interface 305, including: input unit 306, such as keyboard, mouse, etc.; output unit 307, such as various types of monitors, speakers, etc.; storage unit 308, such as disk, optical disk, etc.; and communication unit 309, such as network card, modem, wireless transceiver, etc. Communication unit 309 allows device 300 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0149] The computing unit 301 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 301 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 301 performs the various methods and processes described above, such as information recommendation methods. For example, in some embodiments, the information recommendation method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 308. In some embodiments, part or all of the computer program may be loaded and / or installed on device 300 via ROM 302 and / or communication unit 309. When the computer program is loaded into RAM 303 and executed by the computing unit 301, one or more steps of the information recommendation method described above may be performed. Alternatively, in other embodiments, the computing unit 301 may be configured to perform information recommendation methods by any other suitable means (e.g., by means of firmware).

[0150] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0151] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0152] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0153] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0154] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0155] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0156] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0157] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. An information recommendation method comprising: obtaining N pieces of historical search information matching a target search information input by a user from log data, N being a positive integer; determining a list of synonyms of a word in the N pieces of historical search information; in a case where the N pieces of historical search information include a first piece of historical search information, deleting the first piece of historical search information from the N pieces of historical search information to obtain first candidate recommendation information, the first piece of historical search information including a first word, the list of synonyms of the first word having an intersection with a list of synonyms of a second word, the second word including at least one of the following: a word in the first piece of historical search information, a word in a second piece of historical search information in the N pieces of historical search information corresponding to the first word, and a word in the target search information corresponding to the first word; determining a first core word in M words in the target search information; obtaining a descriptor associated with the first core word in a database, the database storing core words and descriptors in association; concatenating the first core word and the descriptor associated with the first core word to obtain second candidate recommendation information; for information in an intersection of the first candidate recommendation information and the second candidate recommendation information, determining a first recommendation weight of the information in the intersection based on an average of a matching degree with the target search information and a weight corresponding to the descriptor, and a preset first channel weight; for information outside the intersection in the first candidate recommendation information, determining a first recommendation weight of the information outside the intersection based on a matching degree with the target search information and a preset first channel weight; for information outside the intersection in the second candidate recommendation information, determining a second recommendation weight of the information outside the intersection according to a weight of the descriptor and a preset second channel weight; determining target recommendation information from the first candidate recommendation information and the second candidate recommendation information based on the first recommendation weight and the second recommendation weight; and performing information recommendation based on the target recommendation information.

2. The method of claim 1, before the deleting the first piece of historical search information from the N pieces of historical search information to obtain the first candidate recommendation information, the method further comprises at least one of the following: for each piece of historical search information in the N pieces of historical search information, in a case where a list of synonyms of any two words in the historical search information has an intersection, determining the historical search information as the first piece of historical search information; in a case where there is a first target piece of historical search information in the N pieces of historical search information, determining the first target piece of historical search information as the first piece of historical search information, for each word in the first target piece of historical search information, there is a word corresponding to the word in the target search information, and in the first target piece of historical search information and the target search information, a list of synonyms of each pair of corresponding words has an intersection. In the case that N is greater than 1, for each two of the N historical search information, if there is an intersection between the synonym lists of each two corresponding words in the two historical search information, the historical search information of one of the two historical search information is determined as the first historical search information, and the two corresponding words are from the two historical search information respectively.

3. The method of claim 1, wherein before the deleting the first historical search information from the N historical search information to obtain the first candidate recommendation information, the method further comprises: in the case that there is a second target historical search information in the N historical search information, determining the second target historical search information as the first historical search information, and the target search information comprises the second target historical search information.

4. The method of claim 1, wherein, M is greater than 1, and the determining the first core word from the M words comprises: performing category analysis on each word in the M words to obtain category information corresponding to the word; determining at least one candidate word from the M words, each candidate word corresponding to category information intersecting with the category information obtained by performing category analysis on the target search information; determining the first core word from the at least one candidate word.

5. An information recommendation device, comprising: a first acquisition module configured to acquire N historical search information matching a target search information input by a user from log data, N being a positive integer; a first determination module configured to determine synonym lists of words in the N historical search information; a deletion module configured to delete a first historical search information from the N historical search information to obtain a first candidate recommendation information in the case that the N historical search information comprises the first historical search information, the first historical search information comprising a first word, and the synonym list of the first word intersecting with a synonym list of a second word, the second word comprising at least one of the following: a word in the first historical search information, a word corresponding to the first word in a second historical search information in the N historical search information, and a word corresponding to the first word in the target search information; a recommendation module configured to perform information recommendation based on the first candidate recommendation information; the device further comprises: a sixth determination module configured to determine a first core word from M words in the target search information; a second acquisition module configured to acquire a descriptive word associated with the first core word in a database, the database storing core words and descriptive words in association; a splicing module configured to splice the first core word and the descriptive word associated with the first core word to obtain a second candidate recommendation information. The recommendation module is specifically configured to: for information in an intersection of the first candidate recommendation information and the second candidate recommendation information, determine a first recommendation weight of the information in the intersection based on an average value of a matching degree with the target search information and a weight corresponding to a description word, and a preset first channel weight; for information outside the intersection in the first candidate recommendation information, determine a first recommendation weight of the information outside the intersection based on a matching degree with the target search information and a preset first channel weight; for information outside the intersection in the second candidate recommendation information, determine a second recommendation weight of the information outside the intersection according to a weight of the description word and a preset second channel weight; and determine target recommendation information from the first candidate recommendation information and the second candidate recommendation information based on the first recommendation weight and the second recommendation weight; and perform information recommendation based on the target recommendation information.

6. The apparatus of claim 5, further comprising: a second determination module configured to, for each of the N historical search information, determine the historical search information as the first historical search information if a synonym list of any two words in the historical search information has an intersection; a third determination module configured to, if there is a first target historical search information in the N historical search information, determine the first target historical search information as the first historical search information, and for each word in the first target historical search information, determine that there is a word corresponding to the word in the target search information, and that a synonym list of each two corresponding words in the first target historical search information and the target search information has an intersection; a fourth determination module configured to, if N is greater than 1, for each two historical search information in the N historical search information, if a synonym list of each two corresponding words in the two historical search information has an intersection, determine a historical search information of one of the two historical search information as the first historical search information, the two corresponding words in the two historical search information being from the two historical search information respectively.

7. The apparatus of claim 5, further comprising: a fifth determination module configured to, if there is a second target historical search information in the N historical search information, determine the second target historical search information as the first historical search information, and the target search information comprising the second target historical search information.

8. The apparatus of claim 5, wherein, M is greater than 1, and the sixth determination module is specifically configured to: perform category analysis on each word in the M words to obtain category information corresponding to the word; determine at least one candidate word from the M words, each candidate word corresponding to category information having an intersection with category information obtained by performing category analysis on the target search information; determine the first core word from the at least one candidate word.

9. An electronic device, comprising: at least one processor; and a memory connected to the at least one processor in communication; wherein, The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-4.

10. A non-transitory computer readable storage medium having stored thereon computer instructions, wherein, The computer instructions are for causing the computer to perform the method of any one of claims 1-4.

11. A computer program product comprising a computer program which, when executed by a processor, implements the method of any one of claims 1-4.

Images