Information processing method and device, computer readable storage medium and electronic equipment
By calculating the vector of the target user in the virtual social scenario and recommending friends based on the vector, the problem of low accuracy of friend recommendations in the virtual social scenario is solved, and more accurate user matching and recommendation effects are achieved.
Patent Information
- Application Number
- CN202411025009.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-29
- Publication Date
- 2025-01-17
AI Technical Summary
In virtual social scenarios, the accuracy of recommending friends to users is low, and the existing technology has not effectively solved this problem.
By determining the users associated with the target user under multiple social categories in the virtual social scene, obtaining user information of each reference user, calculating the vector of the target user, and determining the target recommended user from the users to be recommended based on the vector, and performing friend recommendations . The specific steps include obtaining the identity information of the reference user, activity information, access information, online duration information, comment information and action track information, performing vector conversion and calculation, using positive feedback coefficients and negative feedback coefficients to adjust the user vector, and combining the word frequency inverse document frequency. The algorithm determines the word segment weight and finally sorts the user according to the similarity.
It improves the accuracy of friend recommendations, can express the preferences and interests of the target users more accurately, thereby improving the user experience, and solving the problem of low accuracy of friend recommendations in virtual social scenarios.
Smart Images
Figure CN119322882A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of artificial intelligence, and in particular to an information processing method, device, computer-readable storage medium and electronic device. Background Art
[0002] In the virtual digital human roaming scene, users carry out activities and socialize in the virtual social scene in the form of digital humans. Some users hope to find people with common characteristics with themselves in this scene. In order to meet user needs, in related technologies, friends of friends are usually directly recommended to users, or friends are recommended to users based on personal information such as the user's area and age, which leads to the problem of low accuracy in recommending friends to users.
[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention
[0004] Embodiments of the present invention provide an information processing method, device, computer-readable storage medium, and electronic device to at least solve the technical problem of low accuracy in recommending friends to users in online virtual social scenarios in related technologies.
[0005] According to one aspect of an embodiment of the present invention, there is provided an information processing method, comprising: determining users associated with a target user under multiple social categories in a virtual social scene, and obtaining multiple reference users, wherein the multiple social categories include: a follow category, a like category, and a not interested category; obtaining user information of each reference user, and obtaining user information sets of multiple reference users, wherein the user information includes identity information of the reference user and activity information of the reference user in the virtual social scene; calculating a target user vector of the target user based on the user information sets of the multiple reference users; determining a target recommended user from the multiple users to be recommended based on the target user vector and user information of the multiple users to be recommended, and making a friend recommendation to the target user based on the target recommended user.
[0006] Furthermore, the information processing method also includes: for each reference user, obtaining the reference user's access information to the virtual social scene, and obtaining the reference user's online time information in the virtual social scene; obtaining the reference user's comment information on the product in the virtual social scene, and obtaining the reference user's action trajectory information in the virtual social scene; determining the reference user's activity information in the virtual social scene based on at least one of the access information, online time information, comment information, and action trajectory information; obtaining the reference user's user information based on the reference user's identity information and activity information, and obtaining a user information set based on the user information of multiple reference users.
[0007] Furthermore, the multiple reference users include a first reference user belonging to the attention category, a second reference user belonging to the like category, and a third reference user belonging to the uninterested category, wherein the information processing method also includes: performing vector conversion on user information of the first reference user in the user information set to obtain a first user vector; performing vector conversion on user information of the second reference user in the user information set to obtain a second user vector; performing vector conversion on user information of the third reference user in the user information set to obtain a third user vector; and calculating a target user vector based on a first positive feedback coefficient corresponding to the first reference user, a second positive feedback coefficient corresponding to the second reference user, a negative feedback coefficient corresponding to the third reference user, and the first user vector, the second user vector, and the third user vector, wherein the first positive feedback coefficient is greater than the second positive feedback coefficient.
[0008] Furthermore, the information processing method also includes: determining structured data information and unstructured data information from the user information of the first reference user; encoding the structured data information to obtain a first sub-vector; for each data information in the unstructured data information, performing word segmentation on the data information to obtain a word segmentation set of the data information; determining a vector representation of the data information based on the weights of the word segments in the word segmentation set in the data information, and determining a second sub-vector based on the vector representation of the data information in the unstructured data information; and obtaining a first user vector based on the first sub-vector and the second sub-vector.
[0009] Furthermore, the information processing method also includes: before determining the vector representation of the data information based on the weight of the segmentation in the segmentation set in the data information, for each segmentation in the segmentation set of the data information, counting the number of target data information containing the segmentation in the virtual social scene to obtain a first value, wherein the target data information is the same as the data type of the data information; counting the total number of target data information in the virtual social scene to obtain a second value; counting the number of occurrences of the segmentation in the data information to obtain a third value; and based on the word frequency inverse document frequency algorithm, calculating the weight of the segmentation in the data information according to the first value, the second value, and the third value.
[0010] Furthermore, the information processing method also includes: performing vector conversion on the user information of each user to be recommended to obtain a vector set of users to be recommended; and determining a target recommended user from multiple users to be recommended based on the similarity between the target user vector and the vector of each user to be recommended in the vector set of users to be recommended.
[0011] Furthermore, the information processing method also includes: sorting the multiple users to be recommended according to the order of similarity from high to low to obtain the sorted multiple users to be recommended; determining the first N users to be recommended among the sorted multiple users to be recommended as target recommended users, where N is a positive integer.
[0012] According to another aspect of an embodiment of the present invention, an information processing device is also provided, including: a first determination module, used to determine users associated with a target user under multiple social categories in a virtual social scene, and obtain multiple reference users, wherein the multiple social categories include: follow category, like category, and not interested category; an acquisition module, used to obtain user information of each reference user, and obtain user information sets of multiple reference users, wherein the user information includes identity information of the reference user and activity information of the reference user in the virtual social scene; a first calculation module, used to calculate a target user vector of the target user based on the user information sets of multiple reference users; a second determination module, used to determine a target recommended user from multiple users to be recommended based on the target user vector and user information of multiple users to be recommended, and make friend recommendations to the target user based on the target recommended user.
[0013] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned information processing method when running.
[0014] According to another aspect of an embodiment of the present invention, an electronic device is also provided, which includes one or more processors; a memory for storing one or more programs, which, when the one or more programs are executed by the one or more processors, enables the one or more processors to run the programs, wherein the programs are configured to execute the above-mentioned information processing method when running.
[0015] In an embodiment of the present invention, a vector of a target user is determined based on information of users associated with the target user in a virtual social scene, and friends are recommended to the target user based on the vector. Multiple reference users are obtained by determining users associated with the target user under multiple social categories in the virtual social scene, and then user information of each reference user is obtained to obtain a user information set of multiple reference users. Then, a target user vector of the target user is calculated based on the user information set of multiple reference users, so that a target recommended user is determined from multiple users to be recommended based on the target user vector and user information of multiple users to be recommended, and a friend recommendation is made to the target user based on the target recommended user. The multiple social categories include: attention category, like category, and uninterested category, and the user information includes identity information of the reference user and activity information of the reference user in the virtual social scene.
[0016] In the above process, by determining the users associated with the target user under multiple social categories in the virtual social scene, multiple reference users are obtained, and the accurate determination of users who are interested in the target user to different degrees and users who are not interested in the target user is achieved. By obtaining the user information sets of multiple reference users and calculating the target user vector of the target user based on the user information sets of multiple reference users, the target user vector of the target user is calculated based on the information that can express the preferences of the target user, thereby improving the accuracy of the determined user vector. By determining the target recommended user from multiple users to be recommended based on the target user vector and the user information of multiple users to be recommended to make friend recommendations, it is achieved to recommend friends to the target user based on the vector that can express the preferences of the target user, thereby improving the accuracy of friend recommendations. In addition, by setting the user information to include the identity information of the reference user and the activity information of the reference user in the virtual social scene, it is achieved to combine the relevant information of the reference user in the virtual social scene to express the preferences of the target user in the virtual social scene, thereby further improving the accuracy of recommending friends to users in the virtual social scene.
[0017] It can be seen that the solution provided in the present application achieves the purpose of determining the vector of the target user based on the information of users associated with the target user in the virtual social scene, and recommending friends to the target user based on the vector, thereby achieving the technical effect of improving the accuracy of friend recommendations, and further solving the technical problem of low accuracy in recommending friends to users in online virtual social scenes in related technologies. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:
[0019] Figure 1 is a schematic diagram of an optional information processing method according to an embodiment of the present invention;
[0020] Figure 2 is a schematic diagram of an optional method for determining a target user vector according to an embodiment of the present invention;
[0021] Figure 3 is a schematic diagram of an optional information processing device according to an embodiment of the present invention;
[0022] Figure 4 is a schematic diagram of an optional electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION
[0023] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.
[0024] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0025] 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, 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 must comply with relevant laws, regulations and standards in the relevant regions, and provide corresponding operation entrances for users to choose to authorize or refuse.
[0026] Example 1
[0027] According to an embodiment of the present invention, an embodiment of an information processing method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.
[0028] Figure 1 is a schematic diagram of an optional information processing method according to an embodiment of the present invention, such as Figure 1 As shown, the method comprises the following steps:
[0029] Step S101, determining users associated with a target user in multiple social categories in a virtual social scene to obtain multiple reference users, wherein the multiple social categories include: a follow category, a like category, and a not interested category.
[0030] Optionally, electronic devices, application systems, servers and other devices can be used as the execution subject of the present application. In this embodiment, the target processing system is used as the execution subject to execute the information processing method. Optionally, the target user is a user to be recommended as a friend in a virtual social scene.
[0031] In some embodiments, the virtual social scene is a social interaction environment built based on virtual reality, augmented reality or other computer-generated technologies. In such a scene, users can communicate, cooperate and interact through virtual characters (such as digital people) instead of their real-life selves. Virtual social scenes vary according to different actual applications. For example, in a financial scene, a virtual social scene may include multiple financial product shops. Users can operate their corresponding digital people to enter the financial product shops in the virtual social scene, purchase financial products in the financial product shops, comment on financial products, etc. Users can also operate their corresponding digital people to add friends, follow, like, mark as uninterested, etc. to other digital people in the virtual social scene, and operate their corresponding digital people to communicate with other digital people in the virtual social scene.
[0032] Optionally, the target processing system may determine the users associated with the target user under multiple social categories in the virtual social scene to obtain multiple reference users. For example, the target user may follow, like, or mark other users' digital people as not interested in the virtual social scene, and the target processing system may determine the users followed, liked, or marked as not interested by the user as reference users.
[0033] Step S102 , obtaining user information of each reference user to obtain a user information set of multiple reference users, wherein the user information includes identity information of the reference user and activity information of the reference user in the virtual social scene.
[0034] Optionally, the reference user's identity information includes but is not limited to office address, job level, years of work experience, gender, etc., and the reference user's activity information in the virtual social scene includes but is not limited to the reference user's access information to the virtual social scene (such as: the time of the last visit, the time range of frequent visits), the reference user's online time information in the virtual social scene (such as: the total online time, the last online time), the reference user's comment information on products in the virtual social scene (such as: financial products), and the reference user's action trajectory information in the virtual social scene. For example, the reference user first enters shop A in the virtual social scene, then enters shop B, and then returns to shop A, then the action trajectory information is "shop A-shop B-shop A".
[0035] Step S103: Calculate a target user vector of the target user based on user information sets of multiple reference users.
[0036] In some embodiments, user information in a user information set of multiple reference users may be vector-converted to obtain a vector corresponding to each user information, and then a target user vector of the target user may be calculated based on the vectors corresponding to the user information of the multiple reference users.
[0037] In some embodiments, the target user vector of the target user can be calculated based on the user information set of multiple reference users based on a text classification algorithm. For example, in a recommendation system, a user portrait vector is constructed by a text classification algorithm, and the correlation between the vector and the features that the user likes is the largest, and the correlation between the vector and the features that the user does not like is the smallest. Further, in the scenario of social friend recommendation, the user's social relationship is analyzed by a text classification algorithm to infer the user's potential interests, and based on this, friends who may be interested are recommended. In a social network, the user's friend relationship can be measured by the proportion of common friends and the interaction situation. The text classification algorithm can be combined with these social relationships and interest similarities to recommend new friends who may be interested to the user. For example, if a user has many common interests with a potential friend and they interact frequently, then this potential friend will be recommended first.
[0038] Step S104 , determining a target recommended user from the multiple users to be recommended according to the target user vector and the user information of the multiple users to be recommended, and making a friend recommendation to the target user based on the target recommended user.
[0039] In some embodiments, the target processing system can perform vector conversion on the user information of the user to be recommended to obtain a vector corresponding to the user information of the user to be recommended, thereby determining a target recommended user from multiple users to be recommended based on the similarity between the target user vector and the vector corresponding to the user information of the user to be recommended, and making friend recommendations to the target user based on the target recommended user.
[0040] Based on the scheme defined in the above steps S101 to S104, it can be known that in an embodiment of the present invention, a vector of a target user is determined according to information of users associated with the target user in a virtual social scene, and friends are recommended to the target user according to the vector. By determining users associated with the target user under multiple social categories in the virtual social scene, multiple reference users are obtained, and then user information of each reference user is obtained to obtain a user information set of multiple reference users. Then, based on the user information sets of multiple reference users, a target user vector of the target user is calculated, so that a target recommended user is determined from multiple users to be recommended according to the target user vector and user information of multiple users to be recommended, and a friend recommendation is made to the target user based on the target recommended user. Among them, the multiple social categories include: attention category, like category, and uninterested category, and the user information includes the identity information of the reference user and the activity information of the reference user in the virtual social scene.
[0041] It is easy to notice that in the above process, by determining the users associated with the target user under multiple social categories in the virtual social scene, multiple reference users are obtained, and the accurate determination of users who are interested in the target user to different degrees and users who are not interested in the target user is achieved. By obtaining the user information sets of multiple reference users and calculating the target user vector of the target user based on the user information sets of multiple reference users, the target user vector of the target user is calculated based on the information that can express the preferences of the target user, thereby improving the accuracy of the determined user vector. By determining the target recommended user from multiple users to be recommended based on the target user vector and the user information of multiple users to be recommended to make friend recommendations, it is achieved to recommend friends to the target user based on the vector that can express the preferences of the target user, thereby improving the accuracy of friend recommendations. In addition, by setting the user information to include the identity information of the reference user and the activity information of the reference user in the virtual social scene, it is achieved to combine the relevant information of the reference user in the virtual social scene to express the preferences of the target user in the virtual social scene, thereby further improving the accuracy of recommending friends to users in the virtual social scene.
[0042] It can be seen that the solution provided in the present application achieves the purpose of determining the vector of the target user based on the information of users associated with the target user in the virtual social scene, and recommending friends to the target user based on the vector, thereby achieving the technical effect of improving the accuracy of friend recommendations, and further solving the technical problem of low accuracy in recommending friends to users in online virtual social scenes in related technologies.
[0043] In an optional embodiment, in the process of obtaining user information of each reference user and obtaining a user information set of multiple reference users, the target processing system can obtain, for each reference user, the reference user's access information to the virtual social scene, and obtain the reference user's online time information in the virtual social scene, obtain the reference user's comment information on products in the virtual social scene, and obtain the reference user's action trajectory information in the virtual social scene, and then determine the reference user's activity information in the virtual social scene based on at least one of the access information, online time information, comment information, and action trajectory information, thereby obtaining the reference user's user information based on the reference user's identity information and activity information, and obtaining a user information set based on the user information of multiple reference users.
[0044] Optionally, the target processing system may be a system for running a virtual social scene, or may be a system capable of communicating with a target server, where the target server is a server running a virtual social scene.
[0045] In some embodiments, when the target processing system is a system for running a virtual social scene, the target processing system can collect and record the above-mentioned access information, online time information, comment information, and action trajectory information of the reference user in the virtual social scene each time the reference user logs into the virtual social scene, so that the aforementioned information can be directly obtained when it is necessary to recommend friends for the reference user.
[0046] In some embodiments, when the target processing system is not a system for running a virtual social scene, the target server can collect and record the above-mentioned access information, online time information, comment information, and action trajectory information of the reference user in the virtual social scene each time the reference user logs into the virtual social scene, so that when the target processing system needs to recommend friends for the reference user, the target processing system interacts with the target server to obtain the aforementioned information.
[0047] Optionally, the reference user's identity information includes but is not limited to office address, job level, years of work experience, gender, etc. The reference user's access information to the virtual social scene may be the time of the last visit, the time range of frequent visits, etc. The reference user's online time information in the virtual social scene may be the total online time, the last online time, etc. The reference user's comment information on products in the virtual social scene (such as financial products) may be "This financial product is very affordable". The reference user's action trajectory information in the virtual social scene may be determined based on the reference user's interactive behavior in the virtual social scene. For example, the reference user first enters shop A in the virtual social scene (equivalent to interacting with shop A), then enters shop B (equivalent to interacting with shop B), and then returns to shop A. The action trajectory information is "shop A-shop B-shop A".
[0048] After determining the access information, online time information, comment information, and action track information, the target processing system may determine at least one of the access information, online time information, comment information, and action track information as the above-mentioned activity information.
[0049] Afterwards, the target processing system may combine the identity information and activity information of the reference user to obtain the user information of the reference user, and further obtain a user information set based on the user information of multiple reference users.
[0050] It should be noted that, through the above process, the richness of the user information in the user information set is improved, so that the accuracy of the target user vector determined subsequently can be effectively improved, thereby improving the accuracy of friend recommendation.
[0051] In an optional embodiment, the plurality of reference users include a first reference user belonging to a concerned category, a second reference user belonging to a liked category, and a third reference user belonging to a not interested category. Figure 2 is a schematic diagram of an optional method for determining a target user vector according to an embodiment of the present invention. Figure 2 As shown, in the process of calculating the target user vector of the target user according to the user information sets of multiple reference users, the target processing system may perform the following steps:
[0052] Step S201 : performing vector conversion on user information of a first reference user in a user information set to obtain a first user vector.
[0053] Step S202: performing vector conversion on the user information of the second reference user in the user information set to obtain a second user vector.
[0054] Step S203: performing vector conversion on the user information of the third reference user in the user information set to obtain a third user vector.
[0055] In some embodiments, the first user vector, the second user vector, and the third user vector of the target processing system are arranged in the same manner.
[0056] In some embodiments, the target processing system can input user information of reference users (i.e., the first reference user, the second reference user, and the third reference user) into a trained neural network model, so that the trained neural network model outputs corresponding user vectors (i.e., the first user vector, the second user vector, and the third user vector).
[0057] In some embodiments, the target processing system may divide the user information of the reference user into structured data information and unstructured data information, and process the structured data information and the unstructured data information using different vector conversion methods to obtain the corresponding user vector.
[0058] Step S204, calculating the target user vector based on the first positive feedback coefficient corresponding to the first reference user, the second positive feedback coefficient corresponding to the second reference user, the negative feedback coefficient corresponding to the third reference user, and the first user vector, the second user vector, and the third user vector, wherein the first positive feedback coefficient is greater than the second positive feedback coefficient.
[0059] In some embodiments, the target processing system may calculate the target user vector using the following formula:
[0060]
[0061] in, represents the target user vector, I r represents the set of first user vectors, I nr represents the set of third user vectors, J r represents the set of second user vectors, represents the i-th first user vector in the set of first user vectors, represents the kth first user vector in the set of third user vectors, represents the lth first user vector in the set of second user vectors, β represents the first positive feedback coefficient, γ represents the negative feedback coefficient, Represents the second positive feedback coefficient.
[0062] In some embodiments, the sizes of the first positive feedback coefficient, the negative feedback coefficient, and the second positive feedback coefficient can be adaptively adjusted according to the number of first reference users, the third reference users, and the second reference users. For example, the first positive feedback coefficient is positively correlated with the number of first reference users, the negative feedback coefficient is positively correlated with the number of third reference users, and the second positive feedback coefficient is positively correlated with the number of second reference users, thereby better reflecting the latest changes in users' interests.
[0063] It should be noted that by assigning different coefficients to user vectors of different social categories and then calculating the target user vector based on the coefficients and the user vector, the accuracy of the determined target user vector is further improved.
[0064] In some embodiments, in the process of performing vector conversion on user information of a first reference user in a user information set to obtain a first user vector, the target processing system can determine structured data information and unstructured data information from the user information of the first reference user, and then encode the structured data information to obtain a first sub-vector, and then for each data information in the unstructured data information, perform word segmentation on the data information to obtain a word segmentation set of the data information, and then determine the vector representation of the data information based on the weights of the word segments in the word segmentation set in the data information, and determine the second sub-vector based on the vector representation of the data information in the unstructured data information, so as to obtain the first user vector based on the first sub-vector and the second sub-vector.
[0065] Optionally, in the target processing system, a correspondence between various types of data information in the user information and structured data information and unstructured data information is preset. The target processing system can distinguish the data information in the user information according to the correspondence. Optionally, identity information, access information, online time information, comment information, and action trajectory information belong to different types of data information.
[0066] For example, for the aforementioned data information of identity information, access information, online time information, comment information and action trajectory information, the data information of identity information, access information and online time information can be determined as structured data information, and the data information of comment information and action trajectory information can be determined as unstructured data information.
[0067] Optionally, for structured data information in user information, the target processing system can encode it using feature engineering to obtain the first sub-vector corresponding to the user information. For example, for data information that is numerical, its corresponding numerical value can be directly determined as a vector value. For example, if there is a data of age, then the value of this data can be directly used as a vector value. For categorical data, one-hot encoding can be used to convert the category into a vector, such as 1 for male gender and 0 for female gender.
[0068] In some embodiments, for each data information in the unstructured data information, since its actual content is text, the data information can be segmented to obtain a segmentation set of the data information. For example, the data information is segmented using the Jieba segmentation method. If the data information is "Shop A-Shop B-Shop A", the segmentation set is "Shop A, Shop B".
[0069] In the process of Jieba word segmentation, the target processing system can scan the word graph to generate a directed acyclic graph of Chinese word segmentation of data information, and then use DP (Dynamic Programming) to find the maximum probability path and find the maximum segmentation combination based on word frequency. After that, the missing words can be calculated based on the Hidden Markov Model (HMM) and Viterbi algorithm, and then the part-of-speech tagging can be done based on the Viterbi algorithm to obtain the final word segmentation result.
[0070] Afterwards, the weights of the segmentations in the segmentation set of the data information in the data information are calculated, and based on the weights of the segmentations in the segmentation set in the data information, the vector representation of the data information is determined. Optionally, the weights of the segmentations in the data information can be calculated based on a word frequency inverse document frequency algorithm, and then the weights of the individual segmentations in the segmentation set are combined to obtain the vector representation of the data information. After obtaining the vector representation of all data information in the unstructured data information of a certain user information, the set of these vector representations can be determined as the second sub-vector corresponding to the user information, thereby combining the first sub-vector and the second sub-vector of the user information to obtain the first user vector.
[0071] For example, the second subvector of a certain user information can be expressed as Here, pn is equivalent to a vector representation of data information. The word set corresponding to the data information can be expressed as T(t1, t2, ... tm). The vector representation corresponding to the data information can be specifically expressed as Among them, w mnrepresents the weight of the mth word tm in data information n. For example, if a piece of data information is a comment, and the specific content of the data information is "This financial product is very affordable", then w mn The meaning can be the weight of the word "affordable" in "This financial product is very affordable."
[0072] It should be noted that by separately vectorizing the structured data information and the unstructured data information in the user information, the accuracy of the determined first user vector can be further improved, thereby improving the accuracy of the user recommendation.
[0073] In an optional embodiment, before determining the vector representation of the data information based on the weight of the segmentation in the segmentation set in the data information, the target processing system can count the number of target data information containing the segmentation in the virtual social scene for each segmentation in the segmentation set of the data information to obtain a first value, then count the total number of target data information in the virtual social scene to obtain a second value, and then count the number of occurrences of the segmentation in the data information to obtain a third value, thereby calculating the weight of the segmentation in the data information based on the first value, the second value, and the third value based on the word frequency inverse document frequency algorithm. Wherein, the target data information is of the same data type as the data information.
[0074] In some embodiments, the target processing system may first calculate the initial weight of the word segmentation in the data information based on the first value, the second value, and the third value based on the word frequency inverse document frequency algorithm, and then normalize the initial weight to obtain the weight of the word segmentation in the data information.
[0075] For example, the target processing system can calculate the initial weight of the word segmentation in the data information through the following formula:
[0076]
[0077] Among them, TF-IDF(tm,p n ) represents the initial weight of the word segment tm in the data information n, TF(tm,p n ) represents the number of occurrences of the participle tm in the data information n, which is also the third value, and N represents the total number of target data information in the virtual social scene, which is also the second value. For example, if the data information is comment information, the sum of the number of comment information (that is, target data information) of all users in the virtual social scene is determined as the second value, and y m It represents the number of target data information containing the segmentation word tm in the virtual social scene, that is, the first value.
[0078] Optionally, after determining the initial weight of the word segmentation in the data information, the target processing system can calculate the weight of the word segmentation in the data information by the following formula:
[0079]
[0080] Among them, w m,n represents the weight of the segmentation tm in the data information n, ts represents the sth segmentation in the data information n, and T represents the segmentation set of the data information n.
[0081] It should be noted that, through the above process, the weight of the word segmentation in the data information is accurately calculated, which can effectively improve the accuracy of the calculated first user vector, and further improve the accuracy of user recommendation.
[0082] In some embodiments, the method for determining the second user vector and the method for determining the third user vector are the same as the method for determining the first user vector, and thus are not described in detail herein.
[0083] In an optional embodiment, in the process of determining a target recommended user from multiple users to be recommended based on the target user vector and user information of multiple users to be recommended, the target processing system can perform vector conversion on the user information of each user to be recommended to obtain a vector set of users to be recommended, thereby determining the target recommended user from multiple users to be recommended based on the similarity between the target user vector and the vector of each user to be recommended in the vector set of users to be recommended.
[0084] Optionally, the process of performing vector conversion on the user information of each to-be-recommended user is the same as the process of performing vector conversion on the user information of the first reference user in the user information set, and therefore will not be described in detail here.
[0085] Optionally, the target processing system may calculate the similarity between the target user vector and the vector of the user to be recommended in the vector set of the user to be recommended based on the cosine similarity algorithm. For example, the similarity is calculated based on the following formula:
[0086]
[0087] in, represents the target user vector, represents the vector of the qth user to be recommended in the vector set of users to be recommended, Represents the similarity between the target user vector and the vector of the qth user to be recommended in the vector set of users to be recommended.
[0088] Afterwards, the target processing system may determine a target recommended user from the plurality of recommended users according to the similarity between the target user vector and the vector of each recommended user in the vector set of recommended users. For example, the recommended user with the highest similarity is determined as the target recommended user, or for another example, the recommended users with the top N similarities are determined as the target recommended users.
[0089] It should be noted that by determining a target recommended user from a plurality of users to be recommended according to vector similarity, the accuracy of the determined target recommended user can be effectively improved.
[0090] In an optional embodiment, in the process of determining a target recommended user from multiple users to be recommended based on the similarity between the target user vector and the vector of each user to be recommended in the vector set of users to be recommended, the target processing system may sort the multiple users to be recommended in order of similarity from high to low to obtain a plurality of sorted users to be recommended, thereby determining the first N users to be recommended among the sorted plurality of users to be recommended as target recommended users, where N is a positive integer.
[0091] Optionally, the above process can also be understood as determining the top N recommended users with the highest similarity as target recommended users.
[0092] After determining the user to be recommended, the target processing system can recommend the target recommended user to the target user in the friend interface in the virtual social scene. Optionally, the target processing system can also send a friend recommendation text message to the target user, and the friend recommendation text message includes the user ID of the target recommended user, thereby realizing friend recommendation.
[0093] It should be noted that by determining the top N recommended users with the highest similarity as target recommended users, the flexibility and accuracy of friend recommendation is improved, and the possibility of target users finding friends is effectively increased, thereby effectively improving user experience.
[0094] It can be seen that the solution provided in the present application achieves the purpose of determining the vector of the target user based on the information of users associated with the target user in the virtual social scene, and recommending friends to the target user based on the vector, thereby achieving the technical effect of improving the accuracy of friend recommendations, and further solving the technical problem of low accuracy in recommending friends to users in online virtual social scenes in related technologies.
[0095] Example 2
[0096] According to an embodiment of the present invention, an embodiment of an information processing device is provided, wherein: Figure 3 is a schematic diagram of an optional information processing device according to an embodiment of the present invention. Figure 3 As shown, the device comprises:
[0097] The first determination module 301 is used to determine users associated with the target user in multiple social categories in the virtual social scene to obtain multiple reference users, wherein the multiple social categories include: a follow category, a like category, and a not interested category;
[0098] An acquisition module 302 is used to acquire user information of each reference user to obtain a user information set of multiple reference users, wherein the user information includes identity information of the reference user and activity information of the reference user in the virtual social scene;
[0099] A calculation module 303, configured to calculate a target user vector of a target user based on user information sets of multiple reference users;
[0100] The second determination module 304 is used to determine a target recommended user from the multiple users to be recommended according to the target user vector and the user information of the multiple users to be recommended, and make a friend recommendation to the target user based on the target recommended user.
[0101] It should be noted that the above-mentioned first determination module 301, acquisition module 302, calculation module 303 and second determination module 304 correspond to steps S101 to S104 in the above-mentioned embodiment, and the examples and application scenarios implemented by the four modules and the corresponding steps are the same, but are not limited to the contents disclosed in the above-mentioned embodiment 1.
[0102] Optionally, the acquisition module also includes: a first acquisition sub-module, used to obtain, for each reference user, the reference user's access information to the virtual social scene, and obtain the reference user's online time information in the virtual social scene; a second acquisition sub-module, used to obtain the reference user's comment information on the product in the virtual social scene, and obtain the reference user's action trajectory information in the virtual social scene; a first determination sub-module, used to determine the reference user's activity information in the virtual social scene based on at least one of the access information, online time information, comment information, and action trajectory information; a second determination sub-module, used to obtain the reference user's user information based on the reference user's identity information and activity information, and obtain a user information set based on the user information of multiple reference users.
[0103] Optionally, the first calculation module also includes: a first processing submodule, used to perform vector conversion on user information of a first reference user in the user information set to obtain a first user vector; a second processing submodule, used to perform vector conversion on user information of a second reference user in the user information set to obtain a second user vector; a third processing submodule, used to perform vector conversion on user information of a third reference user in the user information set to obtain a third user vector; and a calculation submodule, used to calculate a target user vector based on a first positive feedback coefficient corresponding to the first reference user, a second positive feedback coefficient corresponding to the second reference user, a negative feedback coefficient corresponding to the third reference user, and the first user vector, the second user vector, and the third user vector, wherein the first positive feedback coefficient is greater than the second positive feedback coefficient.
[0104] Optionally, the first processing submodule also includes: a first determination unit, used to determine structured data information and unstructured data information from user information of a first reference user; a first processing unit, used to encode the structured data information to obtain a first sub-vector; a second processing unit, used to perform word segmentation on each piece of data information in the unstructured data information to obtain a word segmentation set of the data information; a second determination unit, used to determine the vector representation of the data information based on the weights of the word segments in the word segmentation set in the data information, and determine the second sub-vector based on the vector representation of the data information in the unstructured data information; and a third determination unit, used to obtain a first user vector based on the first sub-vector and the second sub-vector.
[0105] Optionally, the information processing device also includes: a first statistical module, used to count the number of target data information containing the segmentation in the virtual social scene for each segmentation in the segmentation set of the data information, and obtain a first value, wherein the target data information is the same as the data type of the data information; a second statistical module, used to count the total number of target data information in the virtual social scene, and obtain a second value; a third statistical module, used to count the number of occurrences of the segmentation in the data information, and obtain a third value; a second calculation module, used to calculate the weight of the segmentation in the data information based on the first value, the second value, and the third value based on the word frequency inverse document frequency algorithm.
[0106] Optionally, the second determination module also includes: a fourth processing submodule, used to perform vector conversion on the user information of each user to be recommended to obtain a vector set of users to be recommended; a third determination submodule, used to determine a target recommended user from multiple users to be recommended based on the similarity between the target user vector and the vector of each user to be recommended in the vector set of users to be recommended.
[0107] Optionally, the third determination submodule also includes: a sorting unit, used to sort multiple users to be recommended according to the order of similarity from high to low, to obtain multiple sorted users to be recommended; a fourth determination unit, used to determine the first N users to be recommended among the sorted multiple users to be recommended as target recommended users, where N is a positive integer.
[0108] Example 3
[0109] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is further provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned information processing method when running.
[0110] Example 4
[0111] According to another aspect of an embodiment of the present invention, an electronic device is provided, wherein: Figure 4 is a schematic diagram of an optional electronic device according to an embodiment of the present invention, such as Figure 4 As shown, the electronic device includes one or more processors; a memory for storing one or more programs, which, when the one or more programs are executed by the one or more processors, enables the one or more processors to run the programs, wherein the programs are configured to execute the above-mentioned information processing method when running.
[0112] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.
[0113] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0114] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.
[0115] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed over multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.
[0116] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0117] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.
[0118] The above are only preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. An information processing method, characterized in that: include: Determine users associated with the target user in multiple social categories in the virtual social scene to obtain multiple reference users, wherein the multiple social categories include: a follow category, a like category, and a not interested category; Acquire user information of each reference user to obtain a user information set of the multiple reference users, wherein the user information includes identity information of the reference user and activity information of the reference user in the virtual social scene; Calculating a target user vector of the target user according to the user information sets of the multiple reference users; According to the target user vector and user information of multiple users to be recommended, a target recommended user is determined from the multiple users to be recommended, and a friend recommendation is made to the target user based on the target recommended user.
2. The method according to claim 1, characterized in that Obtaining user information of each reference user to obtain a user information set of the multiple reference users includes: For each reference user, obtaining access information of the reference user to the virtual social scene, and obtaining online time information of the reference user in the virtual social scene; Obtaining the review information of the reference user on the product in the virtual social scene, and obtaining the action track information of the reference user in the virtual social scene; Determine the activity information of the reference user in the virtual social scene according to at least one of the access information, the online time information, the comment information, and the action trajectory information; The user information of the reference user is obtained according to the identity information and activity information of the reference user, and the user information set is obtained according to the user information of the multiple reference users.
3. The method according to claim 1, characterized in that: The multiple reference users include a first reference user belonging to the attention category, a second reference user belonging to the like category, and a third reference user belonging to the uninterested category, wherein the target user vector of the target user is calculated according to the user information sets of the multiple reference users, including: Performing vector conversion on user information of a first reference user in the user information set to obtain a first user vector; Performing vector conversion on the user information of a second reference user in the user information set to obtain a second user vector; Performing vector conversion on the user information of a third reference user in the user information set to obtain a third user vector; The target user vector is calculated based on a first positive feedback coefficient corresponding to the first reference user, a second positive feedback coefficient corresponding to the second reference user, a negative feedback coefficient corresponding to the third reference user, and the first user vector, the second user vector, and the third user vector, wherein the first positive feedback coefficient is greater than the second positive feedback coefficient.
4. The method according to claim 3, characterized in that Performing vector conversion on the user information of the first reference user in the user information set to obtain a first user vector includes: determining structured data information and unstructured data information from the user information of the first reference user; Encoding the structured data information to obtain a first sub-vector; For each piece of data information in the unstructured data information, performing word segmentation processing on the data information to obtain a word segmentation set of the data information; Determine a vector representation of the data information according to weights of the segmentations in the segmentation set in the data information, and determine a second sub-vector according to the vector representation of the data information in the unstructured data information; The first user vector is obtained according to the first sub-vector and the second sub-vector.
5. The method according to claim 4, characterized in that Before determining the vector representation of the data information according to the weights of the segmentations in the segmentation set in the data information, the method further includes: For each segmentation in the segmentation set of the data information, counting the number of target data information containing the segmentation in the virtual social scene to obtain a first value, wherein the target data information has the same data type as the data information; Counting the total amount of target data information in the virtual social scene to obtain a second value; Counting the number of occurrences of the word in the data information to obtain a third value; Based on a word frequency inverse document frequency algorithm, the weight of the word segment in the data information is calculated according to the first value, the second value, and the third value.
6. The method according to claim 1, characterized in that Determining a target recommended user from the multiple users to be recommended according to the target user vector and user information of the multiple users to be recommended includes: Performing vector conversion on the user information of each of the to-be-recommended users to obtain a vector set of the to-be-recommended users; The target recommended user is determined from the multiple users to be recommended according to the similarity between the target user vector and the vector of each user to be recommended in the vector set of users to be recommended.
7. The method according to claim 6, characterized in that Determining the target recommended user from the plurality of users to be recommended according to the similarity between the target user vector and the vector of each user to be recommended in the vector set of users to be recommended includes: Sorting the plurality of users to be recommended according to the order of the similarities from high to low to obtain a plurality of sorted users to be recommended; The first N users to be recommended among the sorted multiple users to be recommended are determined as the target recommended users, where N is a positive integer.
8. An information processing device, characterized in that: include: A first determination module is used to determine users associated with the target user in multiple social categories in the virtual social scene to obtain multiple reference users, wherein the multiple social categories include: a follow category, a like category, and a not interested category; An acquisition module, configured to acquire user information of each reference user to obtain a user information set of the plurality of reference users, wherein the user information includes identity information of the reference user and activity information of the reference user in the virtual social scene; A first calculation module, configured to calculate a target user vector of the target user according to the user information sets of the multiple reference users; The second determination module is used to determine a target recommended user from the multiple users to be recommended according to the target user vector and user information of the multiple users to be recommended, and make a friend recommendation to the target user based on the target recommended user.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program is configured to execute the information processing method described in any one of claims 1 to 7 when executed.
10. An electronic device, characterized in that: The electronic device includes one or more processors; A memory for storing one or more programs, which, when the one or more programs are executed by the one or more processors, enables the one or more processors to run the programs, wherein the programs are configured to execute the information processing method described in any one of claims 1 to 7 when running.