Cross-domain preference prediction recommendation method and system based on personality characteristics
By acquiring users' personality trait data and using personality preference association models and historical behavior data to generate recommendation lists, the problems of 'information cocoon' and 'cold start' in traditional recommendation systems are solved. This enables accurate prediction and recommendation of cross-domain interests, improving the novelty and user satisfaction of the recommendation system.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING ZHILUN ROTATION TECHNOLOGY CO LTD
- Filing Date
- 2026-01-26
- Publication Date
- 2026-05-08
AI Technical Summary
Traditional recommendation systems rely on users' historical behavior data, leading to the problems of 'information cocoons' and 'cold start', and are unable to effectively predict users' cross-domain interests and the preferences of new users and new items.
By acquiring users' raw personality data, converting it into personality feature representations, using a personality preference association model to predict the preference scores of items that users have not yet encountered, and combining it with historical behavioral data to generate a recommendation list, dynamically adjusting the weights for deep integration.
It enables accurate prediction of people, information, and items that users may not have encountered but may prefer, expands the boundaries of user interests, improves the novelty and satisfaction of recommendations, and solves the problems of 'information cocoon' and 'cold start'.
Smart Images
Figure CN121996844A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of digital content distribution technology, specifically to a cross-domain preference prediction and recommendation method and system based on personality traits. Background Technology
[0002] With the explosive growth of internet information, personalized recommendation systems have become a core technology in e-commerce, digital content distribution, and social networks. Traditional recommendation systems primarily rely on users' historical behavioral data (e.g., browsing, clicks, purchases, and rating records) to mine and predict preferences. However, traditional recommendation models suffer from two fundamentally related flaws:
[0003] First, there's the "information cocoon" effect: traditional recommendation systems continuously recommend content highly similar to a user's past interests, causing the range of information a user encounters to narrow, gradually trapping them within their existing interests and preferences. This not only limits the possibility of users exploring new fields and expanding their cognitive boundaries, but also reduces users' sense of novelty and satisfaction with recommendation systems in the long run.
[0004] Secondly, there is the "cold start" problem: for new users, traditional recommendation systems cannot generate effective personalized recommendations because they have not yet generated enough historical behavioral data. Similarly, for newly emerging content or products on the platform that lack user interaction data (i.e., "new item cold start"), the system also struggles to accurately reach them.
[0005] In essence, the limitation of existing technologies lies in the fact that their recommendation logic is entirely based on users' explicit behaviors that have already occurred, and they cannot effectively understand and quantify the intrinsic and stable psychological motivations that drive user behavior. Therefore, traditional systems can only mine and repeat users' known interests, and are completely unable to predict users' potential preferences for areas or items they have never encountered before, making them inadequate when dealing with scenarios involving interest expansion and new users / new items.
[0006] Therefore, the industry urgently needs a recommendation technology that can break through the limitations of historical behavioral data, understand users' intrinsic characteristics at a deeper level, and predict cross-domain interests accordingly, so as to fundamentally alleviate the problems of "information cocoon" and "cold start" and thus improve the exploratory capabilities of recommendation systems and the long-term user experience. Summary of the Invention
[0007] To address this, this application provides a cross-domain preference prediction and recommendation method and system based on personality traits to solve the information cocoon and cold start problems existing in the prior art.
[0008] To achieve the above objectives, this application provides the following technical solution:
[0009] Firstly, a cross-domain preference prediction and recommendation method based on personality traits includes:
[0010] Step 1: Obtain the user's raw personality data and historical behavior data, and convert the raw personality data into a personality feature representation;
[0011] Step 2: Based on the personality trait representation, predict the user's preference score for untouched items using a pre-built personality preference association model;
[0012] Step 3: Generate a first recommendation list based on the preference scores;
[0013] Step 4: Generate a second recommendation list based on the historical behavior data;
[0014] Step 5: Dynamically adjust the weights of the first recommendation list and the second recommendation list based on the recommendation scenario and user feedback;
[0015] Step 6: Perform deep fusion of the first recommendation list and the second recommendation list according to the weights, and generate the final recommendation list.
[0016] Preferably, in step 1, when obtaining the user's original personality data, it is obtained from the user's authorized psychological assessment questionnaire, the user's language behavior analysis, and multimodal interaction data.
[0017] Preferably, in step 1, the personality characteristics are represented by personality feature vectors, and psychological theories and data modeling techniques are used when converting the original personality data into personality feature vectors.
[0018] Preferably, in step 2, the personality preference association model is constructed by learning the relationship between personality traits and cross-domain preferences in large-scale sample data through machine learning or statistical analysis algorithms.
[0019] Preferably, in step 2, the preference score includes social preference score, content preference score, and product preference score.
[0020] Preferably, in step 4, a collaborative filtering algorithm or a content recommendation algorithm is used when generating the second recommendation list based on the historical behavior data.
[0021] Secondly, a cross-domain preference prediction and recommendation system based on personality traits includes:
[0022] The data acquisition module is used to acquire the user's raw personality data and historical behavior data, and convert the raw personality data into a personality feature representation;
[0023] The preference score calculation module is used to predict the user's preference score for unvisited items based on the personality trait representation and through a pre-built personality preference association model.
[0024] The first recommendation list generation module is used to generate a first recommendation list based on the preference score;
[0025] The second recommendation list generation module is used to generate a second recommendation list based on the historical behavior data.
[0026] The weight adjustment module is used to dynamically adjust the weights of the first recommendation list and the second recommendation list based on the recommendation scenario and user feedback.
[0027] The fusion recommendation module is used to deeply fuse the first recommendation list and the second recommendation list according to the weights, and generate the final recommendation list.
[0028] Thirdly, a computer device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement steps of a cross-domain preference prediction and recommendation method based on personality traits.
[0029] Fourthly, a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the steps of a cross-domain preference prediction and recommendation method based on personality traits.
[0030] Fifthly, a computer program product includes a computer program or instructions that, when executed by a processor, implement steps of a cross-domain preference prediction and recommendation method based on personality traits.
[0031] Compared with the prior art, this application has at least the following beneficial effects:
[0032] Based on further analysis and research of existing technical problems, this application provides a cross-domain preference prediction and recommendation method based on personality traits. It acquires users' raw personality data and historical behavioral data, converting the raw personality data into personality trait representations. Based on these representations, a pre-constructed personality preference association model predicts users' preference scores for unexplored items. A first recommendation list is generated based on these preference scores; a second recommendation list is generated based on historical behavioral data; the weights of the first and second recommendation lists are dynamically adjusted based on the recommendation scenario and user feedback; and the first and second recommendation lists are deeply integrated according to their weights to generate the final recommendation list. This application, by establishing a user personality preference association model, achieves accurate prediction and recommendation of people, information, and items that users have never encountered but are highly likely to prefer, thereby expanding the boundaries of users' interests, improving the novelty and satisfaction of recommendations, and solving the "information cocoon" and "cold start" problems caused by traditional recommendation methods that rely solely on users' historical behavior. Attached Figure Description
[0033] To more intuitively illustrate the prior art and this application, exemplary drawings are provided below. It should be understood that the specific shapes and structures shown in the drawings should not generally be regarded as limiting conditions for implementing this application; for example, based on the technical concept disclosed in this application and the exemplary drawings, those skilled in the art are able to easily make conventional adjustments or further optimizations to the addition / reduction / classification, specific shapes, positional relationships, connection methods, size ratios, etc. of certain units (components).
[0034] Figure 1 The flowchart of a cross-domain preference prediction and recommendation method based on personality characteristics is provided in Embodiment 1 of this application. Detailed Implementation
[0035] The present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0036] In the description of this application: unless otherwise stated, "a plurality of" means two or more. The terms "first," "second," "third," etc., in this application are intended to distinguish the objects referred to and do not have any special meaning in terms of technical connotation (e.g., they should not be construed as an emphasis on importance or order). Expressions such as "including," "comprising," and "having" also mean "not limited to" (certain units, components, materials, steps, etc.).
[0037] The terms used in this application, such as "upper," "lower," "left," "right," and "middle," are generally used to indicate the general relative positional relationship for the purpose of intuitive understanding by referring to the accompanying drawings, and are not absolute limitations on the positional relationship in the actual product.
[0038] Example 1
[0039] Please see Figure 1 This embodiment provides a cross-domain preference prediction and recommendation method based on personality traits, including:
[0040] S1: Obtain the user's raw personality data and historical behavior data, and convert the raw personality data into a personality trait representation;
[0041] Specifically, this step obtains raw personality data from users through various means, including user-authorized psychological assessment questionnaires, analysis of users' language behavior (e.g., social media writing style, word usage habits), and multimodal interaction data (e.g., facial expressions, voice).
[0042] After obtaining the user's raw personality data, psychological theories (such as the Big Five personality traits, MBTI, etc.) and data modeling techniques are used to transform the raw personality data into a multi-dimensional, quantifiable personality feature vector P = (p1, p2,..., pn). (In practice, it is not limited to using vectors to represent it).
[0043] This step transforms the abstract representation of personality traits into a computable numerical vector. This is the cornerstone of the entire method.
[0044] S2: Based on personality trait representation, predict users' preference scores for unvisited items through a pre-built personality preference association model;
[0045] Specifically, this embodiment pre-establishes a scientific correlation model between personality traits and preferences across different domains, namely, a personality preference correlation model. This is an innovative "knowledge graph." More specifically: the personality preference correlation model is constructed by learning the correlation between personality traits and cross-domain preferences in large-scale sample data through machine learning or statistical analysis algorithms, including:
[0046] Data sources: Collected from multiple sources including psychology research papers, market research reports, and large-scale user sampling surveys;
[0047] Association Learning: Using machine learning algorithms, we analyze the statistically significant preferences exhibited by different personality vector groups on the following aspects:
[0048] (1) Social preferences: What personality traits do people tend to associate with?
[0049] (2) Content preferences: What type (e.g., suspense, comedy, science) and style (e.g., fast-paced, slow-paced narrative) of video, audio and text content do you prefer to consume?
[0050] (3) Product preferences: What design style, brand positioning and function orientation (e.g., practical vs. enjoyment) do you prefer to buy?
[0051] Generate mapping functions: Finally, one or more mapping functions F(p) -> preference_score are generated, thus obtaining the personality preference association model (i.e., the personality-preference association mapping library).
[0052] The personality preference association model constructed in this embodiment takes a user's personality feature vector as input and can predict their preference scores on different items, including social preference scores, content preference scores, and product preference scores.
[0053] S3: Generate a first recommendation list based on preference scores;
[0054] Specifically, this embodiment uses a "personality-preference association mapping library" to predict and score items (especially new content / products) not covered in the user's historical behavior, which can generate a recommendation list List_A, i.e. the first recommendation list, thereby discovering the user's potential new interests.
[0055] S4: Generate a second recommendation list based on historical behavior data;
[0056] Specifically, this step runs traditional collaborative filtering algorithms or content recommendation algorithms to generate a recommendation list List_B, i.e., the second recommendation list, based on the user's historical behavior.
[0057] In this embodiment, steps S3 and S4 generate two recommendation lists in a dual-track parallel manner. The first track is personality prediction, which yields the first recommendation list, and the second track is historical behavior prediction, which yields the second recommendation list.
[0058] S5: Dynamically adjust the weights of the first and second recommendation lists based on the recommendation scenario and user feedback;
[0059] Specifically, this step dynamically adjusts the weights of the two tracks based on the recommendation scenario (e.g., solving cold start issues, exploring new interests) and user feedback. List_A is given higher weight when new users are being added or when breaking down information silos is necessary. The introduction of List_A naturally brings diversity and novelty to the recommendation method.
[0060] S6: Based on the weights, the first recommendation list and the second recommendation list are deeply merged to generate the final recommendation list.
[0061] Specifically, this step deeply integrates traditional recommendations (i.e., historical behavior recommendations) with personality prediction recommendations to generate the final recommendation list.
[0062] This embodiment provides a cross-domain preference prediction and recommendation method based on personality traits. By establishing a user's psychological personality model (i.e., a personality preference association model), it achieves accurate prediction and recommendation of people, information, and items that the user has never encountered but is highly likely to prefer. This expands the user's interest boundaries, improves the novelty and satisfaction of recommendations, and solves the "information cocoon" and "cold start" problems caused by traditional recommendation methods that rely solely on the user's historical behavior.
[0063] Example 2
[0064] This embodiment provides a cross-domain preference prediction and recommendation system based on personality traits, including:
[0065] The data acquisition module is used to acquire the user's raw personality data and historical behavior data, and convert the raw personality data into a personality feature representation;
[0066] The preference score calculation module is used to predict the user's preference score for unvisited items based on the personality trait representation and through a pre-built personality preference association model.
[0067] The first recommendation list generation module is used to generate a first recommendation list based on the preference score;
[0068] The second recommendation list generation module is used to generate a second recommendation list based on the historical behavior data.
[0069] The weight adjustment module is used to dynamically adjust the weights of the first recommendation list and the second recommendation list based on the recommendation scenario and user feedback.
[0070] The fusion recommendation module is used to deeply fuse the first recommendation list and the second recommendation list according to the weights, and generate the final recommendation list.
[0071] For details on the specific implementation of each module in a cross-domain preference prediction and recommendation system based on personality characteristics, please refer to the above description of the limitations of a cross-domain preference prediction and recommendation method based on personality characteristics, which will not be repeated here.
[0072] Example 3
[0073] This embodiment provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of a cross-domain preference prediction and recommendation method based on personality characteristics.
[0074] Example 4
[0075] This embodiment provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of a cross-domain preference prediction and recommendation method based on personality traits.
[0076] Example 5
[0077] This embodiment provides a computer program product, including a computer program or instructions, which, when executed by a processor, implements the steps of a cross-domain preference prediction and recommendation method based on personality traits.
[0078] The technical features of the above embodiments can be combined in any way (as long as there is no contradiction in the combination of these technical features). For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described; these embodiments not explicitly written should also be considered to be within the scope of this specification.
Claims
1. A cross-domain preference prediction and recommendation method based on personality traits, characterized in that, include: Step 1: Obtain the user's raw personality data and historical behavior data, and convert the raw personality data into a personality feature representation; Step 2: Based on the personality trait representation, predict the user's preference score for untouched items using a pre-built personality preference association model; Step 3: Generate a first recommendation list based on the preference scores; Step 4: Generate a second recommendation list based on the historical behavior data; Step 5: Dynamically adjust the weights of the first recommendation list and the second recommendation list based on the recommendation scenario and user feedback; Step 6: Perform deep fusion of the first recommendation list and the second recommendation list according to the weights, and generate the final recommendation list.
2. The cross-domain preference prediction and recommendation method based on personality traits according to claim 1, characterized in that, In step 1, when obtaining the user's original personality data, it is obtained from the user's authorized psychological assessment questionnaire, the user's language behavior analysis, and multimodal interaction data.
3. The cross-domain preference prediction and recommendation method based on personality traits according to claim 1, characterized in that, In step 1, the personality characteristics are represented using personality feature vectors. Psychological theories and data modeling techniques are used when converting the original personality data into personality feature vectors.
4. The cross-domain preference prediction and recommendation method based on personality traits according to claim 1, characterized in that, In step 2, the personality preference association model is constructed by learning the relationship between personality traits and cross-domain preferences in large-scale sample data through machine learning or statistical analysis algorithms.
5. The cross-domain preference prediction and recommendation method based on personality traits according to claim 1, characterized in that, In step 2, the preference scores include social preference scores, content preference scores, and product preference scores.
6. The cross-domain preference prediction and recommendation method based on personality traits according to claim 1, characterized in that, In step 4, a collaborative filtering algorithm or a content recommendation algorithm is used to generate the second recommendation list based on the historical behavior data.
7. A cross-domain preference prediction and recommendation system based on personality traits, characterized in that, include: The data acquisition module is used to acquire the user's raw personality data and historical behavior data, and convert the raw personality data into a personality feature representation; The preference score calculation module is used to predict the user's preference score for unvisited items based on the personality trait representation and through a pre-built personality preference association model. The first recommendation list generation module is used to generate a first recommendation list based on the preference score; The second recommendation list generation module is used to generate a second recommendation list based on the historical behavior data. The weight adjustment module is used to dynamically adjust the weights of the first recommendation list and the second recommendation list based on the recommendation scenario and user feedback. The fusion recommendation module is used to deeply fuse the first recommendation list and the second recommendation list according to the weights, and generate the final recommendation list.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.
10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 6.