Retrieval result display method and system

Abstract

The invention discloses a retrieval result display method and system, and relates to the technical field of data processing.The method comprises the steps that a retrieval result tree containing hierarchical nodes is constructed and displayed on a superficial layer of a terminal interface, and an expandable mark is set; collecting user browsing behaviors through sight tracking, and calculating attention popularity; recognizing a retrieval intention in combination with the popularity, the result tree and the user portrait; and generating a deep display instruction according to the retrieval intention, controlling an interface to display a deeper result, and realizing personalized information presentation. Therefore, the technical effects of accurate intention recognition, active information pushing and personalized and efficient retrieval result display are achieved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a search result display method and system. Background Art

[0002] With the rapid development of information technology, search result display methods have been widely used on various terminal devices. On terminal devices with limited space, how to effectively display search results so that users can quickly and accurately obtain the required information has become an urgent problem to be solved.

[0003] Traditional search result display methods have the following technical shortcomings: First, they are unable to keenly capture users' immediate changes in interest and cannot predict user needs in advance. This is because traditional methods mainly rely on users' explicit operations, such as clicking or staying for a long time, which often have certain lags and uncertainties and cannot reflect users' true intentions in a timely manner; second, traditional search result display methods lack flexibility and dynamism, and cannot adapt well to users' changing needs for information depth at different browsing stages. For example, within a limited screen space, it is difficult for users to conveniently adjust the depth and scope of information display according to their needs, which can easily cause confusion or permanent changes in the main interface layout; third, existing technologies find it difficult to comprehensively consider users' long-term interests and short-term concerns without increasing the user's additional operational burden, so as to achieve accurate positioning of the user's current search items, and thus cannot provide truly personalized search result display. Summary of the Invention

[0004] The present invention provides a search result display method and system to solve the technical problems in the existing technology that the display method lacks flexibility and dynamism and is difficult to accurately match users' long-term interests and short-term concerns, and achieves the technical effects of accurate intent recognition and active information push, personalized and efficient search result display.

[0005] In a first aspect, the present invention provides a search result display method, wherein the search result display method comprises:

[0006] The search results are obtained and a search result tree including hierarchical nodes is constructed accordingly, and the search result tree is shallowly displayed on the target display interface of the terminal device and marked as expandable.

[0007] The target user's browsing sight behavior data is collected in real time through an eye tracking device, and a shallow display attention heat set is calculated based on the browsing sight behavior data.

[0008] Screening and matching are performed based on the attention heat set, the search result tree and the user portrait to identify the real-time search intention of the target user.

[0009] A corresponding deep display instruction is generated based on the real-time search intention, and the deep display instruction is used to control the target display interface to perform deep display of the search result tree.

[0010] In a feasible implementation, the search results are obtained and a search result tree including hierarchical nodes is constructed accordingly, and the search result tree is shallowly displayed on the target display interface of the terminal device and is marked as expandable, including:

[0011] A tree display structure is generated according to the association strength and semantic hierarchical relationship of the search results, and is output as the search result tree, wherein the search result tree includes at least one shallow level node and at least one deep level node.

[0012] The expandability of the shallow-level node under which the deep-level node is subordinate is marked as expandable.

[0013] According to the expandability mark structure and the relevance between the shallow level nodes and the search content, the visual content of the shallow level nodes is presented on the target display interface.

[0014] In one feasible implementation, the target user's browsing behavior data is collected in real time through an eye tracking device, including:

[0015] Aligning the target display interface with a timestamp of the gaze tracking device, wherein the gaze tracking device is integrated into the terminal device.

[0016] When the shallow presentation begins, the eye tracking device is synchronously activated to perform eye tracking on the target user.

[0017] The browsing sight behavior data is generated according to the eye tracking result, wherein the browsing sight behavior data includes the sight dwell time, sight trajectory path and number of return glances in each shallow level node area in the target display interface.

[0018] In a feasible implementation, calculating a shallow display attention heat set based on the browsing sight behavior data includes:

[0019] The browsing sight behavior data is preprocessed, and the preprocessing includes normalization processing and dimensionless conversion.

[0020] Confidence analysis is performed on the pre-processed browsing sight behavior data to determine corresponding confidence factors and generate a modified weight set.

[0021] The pre-processed browsing sight behavior data is weightedly calculated in combination with the modified weight set and the preset data fusion weight parameter set to generate attention heat values ​​corresponding to multiple shallow-level nodes to constitute the attention heat set.

[0022] In a feasible implementation, screening and matching is performed based on the attention heat set, the search result tree, and the user profile to identify the real-time search intention of the target user, including:

[0023] User portrait features are extracted from the target user's long-term historical search records, wherein the user portrait features include keyword preference features and high-frequency topic features.

[0024] Traverse the shallow level nodes in the search result tree and extract corresponding node semantic features.

[0025] Based on the user portrait features, the node semantic features are traversed to perform feature similarity matching calculation to obtain the portrait matching degree.

[0026] The portrait matching degree is compared with the portrait matching degree constraint, and the corresponding attention heat value is compared with the attention heat value constraint.

[0027] If there is a portrait matching degree that satisfies the portrait matching degree constraint, and the attention heat value satisfies the attention heat value constraint, then the corresponding shallow level node is determined as the real-time retrieval intention.

[0028] In a feasible implementation, generating corresponding deep display instructions based on the real-time search intent, and controlling the target display interface to perform deep display of the search result tree through the deep display instructions, includes:

[0029] According to the distribution of the attention heat concentration, the display layout of the target display interface is optimized, and the shallow level nodes corresponding to low heat values ​​are folded.

[0030] Based on the optimized display layout, preloading and prerendering are performed on the deep-level nodes under the shallow-level nodes pointed to by the real-time retrieval intention.

[0031] The real-time sight focus of the target user is continuously detected by the sight tracking device. When the real-time sight focus is focused on any shallow-level node, the content of the corresponding deep-level node is displayed in a floating form.

[0032] In a feasible implementation, the method further includes:

[0033] An expanded search based on the user portrait is performed based on deep hierarchical nodes to obtain deep expanded search results.

[0034] The association between the deep expanded search result and the deep level nodes is evaluated, and the deep expanded search result is serialized according to the association evaluation result.

[0035] The deep expanded search results are updated to the search result tree based on the serialization results and displayed.

[0036] In a second aspect, the present invention further provides a search result display system, wherein the search result display system comprises:

[0037] The shallow display module is used to obtain the search results and construct a search result tree including hierarchical nodes, and perform shallow display of the search result tree on the target display interface of the terminal device and perform expandability marking.

[0038] The sight collection and heat calculation module is used to collect the target user's browsing sight behavior data in real time through the sight tracking device, and calculate the shallow display attention heat set based on the browsing sight behavior data.

[0039] The intention recognition module is used to screen and match the attention heat set, the search result tree and the user portrait to identify the real-time search intention of the target user.

[0040] The deep display control module is used to generate corresponding deep display instructions based on the real-time search intention, and control the target display interface to perform deep display of the search result tree through the deep display instructions.

[0041] The present invention discloses a search result display method and system, comprising: after obtaining the search results, constructing a search result tree including hierarchical nodes, and displaying the result tree in a shallow form on a target display interface of a terminal device, while adding marks to nodes that can be further expanded; using an eye tracking device to collect the user's browsing sight behavior data in real time, and calculating the attention heat of each node in the shallow display based on the data; screening and matching the nodes in combination with the attention heat, the search result tree structure and the user portrait, and identifying the user's real-time search intention; generating a deep display instruction according to the identified real-time search intention, and driving the target display interface to display the corresponding node in depth through the instruction. The search result display method and system disclosed by the present invention solve the technical problems of lack of flexibility and dynamism in the display method and difficulty in accurately matching the user's long-term interests and short-term concerns, and achieves the technical effects of accurate intention recognition and active information push, and personalized and efficient search result display. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 The figure is a flow chart of a search result display method of the present invention.

[0043] Figure 2 The figure is a structural diagram of a search result display system of the present invention.

[0044] Description of reference numerals:

[0045] Shallow display module 11, sight acquisition and heat calculation module 12, intention recognition module 13, deep display control module 14. DETAILED DESCRIPTION

[0046] The above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods of the specification to better understand the above technical solution. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments used only to explain the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In addition, it should be noted that, for the convenience of description, only the parts related to the present invention, rather than all, are shown in the drawings.

[0047] Example 1, as Figure 1 The figure is a flow chart of a search result display method of the present invention, wherein the search result display method includes:

[0048] S100: Acquire search results and construct a search result tree including hierarchical nodes accordingly, and perform a shallow display of the search result tree on a target display interface of a terminal device, and perform an expandability mark.

[0049] Specifically, a search result tree is a hierarchical data structure used to organize and present search results in layers. The tree organizes search results into layers based on logical relationships or importance, using hierarchical nodes. Top-level nodes represent the main search topic or category, while sub-nodes contain more specific or related search results.

[0050] Specifically, shallow display refers to displaying only the top-level or partial-level nodes of the search result tree on the terminal device's display interface to provide an overview and avoid information overload. Expandability markers are used to indicate that top-level or partial-level nodes can be expanded to view more detailed content.

[0051] Through the above process, we achieve efficient organization and preliminary hierarchical display of a large number of search results. This allows users to quickly access the overall structure and key information of the search results, avoids the information overload problem in traditional display methods, effectively optimizes display space, and ensures a clear and concise interface.

[0052] In some embodiments, obtaining the search results and correspondingly constructing a search result tree including hierarchical nodes, and shallowly displaying the search result tree on a target display interface of a terminal device and marking it as expandable include:

[0053] Based on the association strength and semantic hierarchical relationship of the search results, a tree-like display structure is generated and output as the search result tree, wherein the search result tree includes at least one shallow-level node and at least one deep-level node; the expandability of the shallow-level node under which the deep-level node is subordinate is marked as expandable; based on the expandability marking structure and the association between the shallow-level node and the search content, the visual content of the shallow-level node is presented on the target display interface.

[0054] Specifically, the search result tree is a hierarchical data structure used to organize search results for easy subsequent presentation. Hierarchical nodes represent the different levels and categories of search results. Shallow presentation refers to displaying only the top level or a portion of the hierarchical nodes of the search result tree on the target display interface of the terminal device. Expandability marks these nodes, prompting users to further expand them to view detailed content.

[0055] Specifically, the association strength refers to the degree of correlation between the retrieval results and the retrieval target, the semantic hierarchical relationship refers to the semantic hierarchy or correlation level of the retrieval results, the shallow level nodes and deep level nodes represent the different levels of depth of the retrieval results respectively, the shallow level nodes are used to display them first on the limited target display interface to improve the efficiency of information overview, and the deep level nodes are presented after the user operation is expanded to achieve a step-by-step in-depth display of information.

[0056] Specifically, first, based on the search content entered by the user, a set of related search results is obtained, and semantic analysis and correlation calculation are performed on each result in the set to construct a tree structure representing the semantic hierarchical relationship between the results. The semantic hierarchical relationship can be determined based on keyword co-occurrence, topic model clustering or knowledge graph path generation.

[0057] Subsequently, a search result tree is generated based on the tree structure, and nodes with associated aggregation features are divided into shallow-level nodes, and the more specific results under them are classified as deep-level nodes. Next, each shallow node under the jurisdiction of a deep-level node is marked as expandable to prompt the user that there is a richer information level under the node. Finally, based on the expandability marking structure and the association priority between the shallow-level nodes and the original search content, the visual content of the shallow-level nodes is selectively presented on the target display interface. For example, the visual content may include node titles, summary information, icons or preview images, adapted to the display capabilities and interaction methods of the terminal device.

[0058] For example, taking the in-vehicle HUD as an example, when the user searches for information about nearby restaurants through the in-vehicle system while driving, a large number of restaurant-related search results are first obtained from the database, including different types of restaurants (such as Chinese restaurants, Western restaurants), user reviews, distance and other information. Then, based on the correlation strength of the search results (such as the degree of matching with the user's historical preferences or the current time) and the semantic hierarchical relationship (such as restaurant type, cuisine classification), a tree display structure is generated and output as a search result tree. The top-level node of the search result tree may be a nearby restaurant, and the shallow-level nodes under it include Chinese restaurants, Western restaurants, etc. These shallow-level nodes may have more detailed deep-level nodes, such as Sichuan restaurants, Italian restaurants, etc. The expandability of these shallow-level nodes (such as "Chinese restaurant") that have deep-level nodes under them can be marked as expandable, and based on the expandability mark structure and the correlation between the shallow-level nodes and the search content (restaurant) (such as the distance from the user's current location, the level of user evaluation, etc.), the visual content of the shallow-level nodes can be presented on the target display interface of the in-vehicle HUD. For example, options such as Chinese restaurant and Western restaurant can be displayed in the form of simple icons or text, and the expandability mark can be used to prompt the user to view more detailed information.

[0059] Through this process, complex search results can be displayed in a structured, hierarchical, and controllable manner on devices with limited display space, avoiding information overload while enhancing users' overall understanding of search results and their ability to delve deeper into specific areas. This approach, particularly in scenarios with limited display space, such as AR glasses, in-vehicle HUDs, or smartwatches, prioritizes shallow information and marks expandable nodes, guiding users to gradually explore deeper content, effectively improving search interaction efficiency and user experience.

[0060] S200: Collecting the target user's browsing sight behavior data in real time through an eye tracking device, and calculating a shallowly displayed attention heat set based on the browsing sight behavior data.

[0061] Specifically, browsing gaze behavior data refers to interactive behavior information such as the user's gaze movement trajectory, gaze point location, dwell time, and browsing order on the target display interface, collected in real time by gaze tracking equipment. It is used to reflect the user's attention to different content areas in the current interface. Among them, the gaze tracking device can be an infrared eye tracker integrated into AR glasses, in-vehicle HUDs, or wearable display terminals, a camera-based gaze tracking module, or an image recognition-based gaze estimation system, which has the ability to dynamically collect the user's gaze direction and gaze area at high frequency.

[0062] Through this process, we can obtain the user's attention to each shallow-level node in real time, providing a basis for subsequent personalized information display and deep-level display decisions. This allows the system to dynamically adjust information display strategies based on the user's actual focus, improving information acquisition efficiency and user experience.

[0063] In some embodiments, the target user's browsing gaze behavior data is collected in real time through an eye tracking device, including:

[0064] Align the target display interface with the timestamp of the gaze tracking device, wherein the gaze tracking device is integrated into the terminal device; when the shallow display starts, synchronously activate the gaze tracking device to perform eye tracking on the target user; generate the browsing gaze behavior data according to the eye tracking result, wherein the browsing gaze behavior data includes the gaze dwell time, gaze trajectory path and number of return glances in each shallow level node area in the target display interface.

[0065] Specifically, first, before the shallow display is started, the target display interface of the terminal device and the integrated gaze tracking device are timestamp-aligned to ensure the timing consistency of the two in data collection and interface rendering. Subsequently, at the same time as the shallow level node display begins, the gaze tracking device is synchronously activated to obtain the target user's eye movement information in real time. Then, based on the eye tracking results, the user's gaze area in the target display interface is identified, and the area is spatially mapped with the corresponding shallow level node to generate browsing gaze behavior data. The browsing gaze behavior data includes the gaze dwell time on each shallow node area (used to measure the user's attention intensity), the gaze trajectory path (used to analyze the user's visual browsing order and jump mode), and the number of return glances (used to reflect the user's repeated attention behavior to a certain node, thereby inferring their information interest or difficulty of understanding).

[0066] This process automatically collects and structures user visual interaction data without burdening users with explicit operations. This provides critical support for generating attention hotspots, dynamically adjusting the display order of nodes, and triggering content recommendations or prompts. This system is particularly suitable for devices with limited display space but high interaction frequency. It can significantly improve the personalization of information display and interactive adaptability, and enhance the system's ability to perceive user intent and respond efficiently.

[0067] In some embodiments, calculating a shallowly displayed attention heat set based on the browsing sight behavior data includes:

[0068] The browsing sight behavior data is preprocessed, and the preprocessing includes normalization and dimensionless conversion; confidence analysis is performed on the preprocessed browsing sight behavior data to determine the corresponding confidence factor and generate a revised weight set; weighted calculation is performed on the preprocessed browsing sight behavior data in combination with the revised weight set and a preset data fusion weight parameter set to generate attention heat values ​​corresponding to multiple shallow-level nodes to constitute the attention heat set.

[0069] Specifically, the collected browsing behavior data is first standardized to improve the stability and adaptability of subsequent calculations. This includes normalization, which normalizes metrics like gaze dwell time, number of return glances, and track length to a uniform range; and dimensionless conversion, which eliminates unit differences between metrics of different dimensions through feature scaling or ratio conversion, ensuring comparability in weighted calculations.

[0070] Specifically, a confidence analysis is then performed on the pre-processed browsing gaze behavior data to assess the credibility of each data indicator and determine the confidence factor corresponding to each indicator. Preferably, this analysis can be based on factors such as the continuity of user browsing behavior, data volatility, device acquisition quality, or the sparseness of gaze in a specific area. Based on the above confidence factors, a modified weight set can be generated to adjust the indicator weights, thereby enhancing the responsiveness to high-confidence data and suppressing the interference of abnormal or low-confidence data on the results.

[0071] Furthermore, the modified weight set is fused with the system's preset data fusion weight parameter set, and a weighted summation or multi-dimensional fusion calculation is performed on the normalized gaze behavior data to obtain the attention heat values ​​corresponding to multiple shallow-level nodes. The data fusion weight parameter set is dynamically adjusted based on different application scenarios, user profiles, or historical behavior characteristics to meet the needs of personalized attention modeling. For example, the importance of gaze dwell time, gaze trajectory path, and number of return glances is determined based on the user profile, and the corresponding data fusion weight parameter set is defined.

[0072] Finally, the attention heat values ​​of all shallow-level nodes are aggregated to form a complete attention heat set, which is used to reflect the relative attention level of users to the content of each node in the current shallow display interface.

[0073] Through the above steps, we can build an attention heat model (i.e., attention heat set) that reflects the individualized attention distribution based on the precise collection of user eye movement data, providing quantitative support for subsequent modules such as shallow node sorting optimization, display rearrangement, interactive recommendation, or deep information prompts.

[0074] S300: Screening and matching are performed based on the attention heat set, the search result tree and the user portrait to identify the real-time search intention of the target user.

[0075] Specifically, real-time search intent dynamically identifies the search targets that a user is most likely to be interested in or seeking based on their current behavior, historical preferences, and content attention patterns. Optionally, this intent is represented as a semantic vector, keyword set, or topic tag, and serves as the core basis for subsequent information filtering, display optimization, and content push.

[0076] In some embodiments, screening and matching are performed based on the attention heat set, the search result tree, and the user profile to identify the real-time search intention of the target user, including:

[0077] Extract user portrait features from the target user's long-term historical search records, wherein the user portrait features include keyword preference features and high-frequency topic features; traverse the shallow-level nodes in the search result tree to extract corresponding node semantic features; based on the user portrait features, traverse the node semantic features to perform feature similarity matching calculation to obtain a portrait matching degree; compare the portrait matching degree with the portrait matching degree constraint, and compare the corresponding attention heat value with the attention heat value constraint; if there is a portrait matching degree that satisfies the portrait matching degree constraint, and the attention heat value satisfies the attention heat value constraint, then determine the corresponding shallow-level node as the real-time search intention.

[0078] Specifically, real-time search intent recognition dynamically determines a user's current needs based on multi-source data fusion. Its core lies in the combined screening and matching of user profile features, the search result tree structure, and attention-focused data sets to accurately locate the user's most relevant interests. This process ensures the accuracy and real-time nature of intent recognition through multi-dimensional feature extraction, semantic matching, and threshold determination.

[0079] Specifically, first, user portrait features are extracted from the target user's long-term historical search records, including keyword preference features and high-frequency topic features. Among them, keyword preference features refer to the set of keywords frequently used by users in past search behaviors, reflecting their stability interest dimension, and high-frequency topic features are the distribution of users' long-term attention topics extracted through topic modeling algorithms (such as LDA or TF-IDF), reflecting their semantic focus. Subsequently, all shallow-level nodes in the current search result tree are traversed, and the pre-trained semantic model is called for each node to extract the node semantic features. The semantic features can include vectorized representations of node title, summary, label and other information for subsequent similarity calculations.

[0080] Specifically, a feature similarity matching calculation (e.g., using cosine similarity, Euclidean distance, or BERT semantic embedding similarity) is then performed based on the extracted user profile features and node semantic features to calculate the profile matching degree of each node with the user profile. This profile matching degree is then compared with the preset profile matching degree constraints to filter out low-correlation nodes. At the same time, the attention value corresponding to each node in the attention heat set is compared with the set attention heat value constraints to ensure that the selected nodes have strong semantic relevance and high current visual attention of the user.

[0081] Furthermore, if there is a shallow-level node that satisfies both the portrait matching degree constraint and the attention heat value constraint, then the node is determined as one of the real-time retrieval intentions of the target user.

[0082] This process, based on the integration of a user's long-term interest profile and short-term visual attention, accurately identifies search intent that is both personalized and highly relevant to the user's current browsing behavior. This serves as the user's most likely search target at the current stage, effectively perceiving the user's real-time search intent. This helps improve the accuracy and personalization of search results, reduces the difficulty users face in selecting results when faced with a large amount of information, and ultimately improves information acquisition efficiency and user experience.

[0083] S400: Generate a corresponding deep display instruction based on the real-time search intention, and control the target display interface to perform deep display of the search result tree through the deep display instruction.

[0084] Specifically, the deep display instruction is a control signal or operation command generated based on the real-time retrieval intention, which is used to trigger the display interface to switch from the shallow display mode to the deep display mode, and specify the detailed content to be displayed. By executing the deep display instruction, the display interface of the terminal device can display more detailed and specific retrieval result information related to the user's real-time retrieval intention.

[0085] Exemplarily, the identified real-time search intent is matched with a display strategy template to determine the display strategy corresponding to the current intent, including the deep node path to be displayed, the recommended expansion range, and priority ranking. The display strategy template is a preset display strategy template corresponding to multiple types of search intent, and includes parameters such as the expansion rules for deep nodes, display levels, content granularity, and interaction methods.

[0086] Exemplarily, a deep display instruction is constructed based on the matched display strategy, which may include: node path identifier of the target search result tree; display level depth; display content type (such as text, image, chart, etc.); node sorting weight; animation or transition effect parameters, etc.

[0087] For example, the generated deep display instruction is sent to the front-end display module or client interface, driving it to expand and display the specified search result tree branch in the current interface. The display process can adopt progressive loading, partial refresh or dynamic re-arrangement to ensure the smoothness and responsiveness of the display effect.

[0088] Through the above process, we can achieve an efficient transition from user intent recognition to information display, providing users with personalized and detailed information. The generation and execution of deep display instructions ensures the accuracy and timeliness of information display, allowing users to quickly obtain content that is highly relevant to their needs.

[0089] In some embodiments, generating a corresponding deep display instruction based on the real-time search intent, and controlling the target display interface to perform a deep display of the search result tree through the deep display instruction, includes:

[0090] According to the distribution of the concentrated attention heat, the display layout of the target display interface is optimized, and the shallow-level nodes corresponding to low heat values ​​are folded; based on the optimized display layout, the deep-level nodes under the shallow-level nodes pointed to by the real-time retrieval intention are preloaded and pre-rendered; the real-time sight focus of the target user is continuously detected by the sight tracking device, and when the real-time sight focus is focused on any shallow-level node, the content of the corresponding deep-level node is displayed in a floating form.

[0091] Specifically, first, the display layout of the target display interface is dynamically optimized according to the heat distribution of the attention heat concentration; among them, the shallow hierarchical nodes whose attention heat values ​​are lower than the preset threshold are folded to reduce redundant information in the interface and improve the user's visual focus efficiency; for example, in the shallow display, there are three nodes of coffee shop, gas station, and restaurant, and the corresponding heat values ​​are 80, 30, and 45 respectively. The two low-heat nodes of gas station and restaurant will be folded, and only the display of the coffee shop node will be retained.

[0092] Specifically, based on the optimized display layout, the shallow-level nodes pointed to by the real-time retrieval intention are identified, and preloading and prerendering operations are performed on the deep-level nodes under them, so as to shorten the user interaction response time and improve the fluency and real-time performance of the deep content display; for example, the relevant information (such as location, evaluation) of the deep-level nodes (such as chain coffee shops, independent coffee shops, etc.) under the coffee shop node is obtained in advance, and preloading and interface prerendering are performed in the background.

[0093] Furthermore, the real-time eye focus position of the target user is continuously monitored through the eye tracking device; when it is detected that the user's eye is focused on any shallow-level node, the floating display mechanism is triggered based on the corresponding deep-level display instruction, and the deep-level node content under the shallow node (i.e., the result of the aforementioned preloading and interface pre-rendering) is dynamically presented, wherein the floating display can be implemented in the form of a floating window, an expanded card, or a dynamic pop-up window to enhance the user's interactive experience and information acquisition efficiency. For example, when the eye tracking device detects that the user's real-time eye focus is focused on a coffee shop node, the detailed content of the chain coffee shop and the independent coffee shop is displayed in a floating form, such as displaying a translucent floating card on the HUD interface, listing the names and distances of several highly rated coffee shops nearby.

[0094] This process not only optimizes the layout of the target display interface and reduces information clutter, but also improves responsiveness through preloading and prerendering, allowing users to experience a smoother browsing experience when browsing content of interest. Furthermore, floating display can provide users with more detailed information without disrupting the main interface structure, making information presentation more intelligent and personalized, thereby better meeting users' immediate needs and improving information acquisition efficiency.

[0095] In some embodiments, further comprising:

[0096] Perform an extended search based on the user portrait based on the deep-level nodes to obtain deep-level extended search results; evaluate the correlation between the deep-level extended search results and the deep-level nodes, and serialize the deep-level extended search results according to the correlation evaluation results; update the deep-level extended search results to the search result tree based on the serialization results, and display them.

[0097] Specifically, expanded search is a further search operation based on deeper nodes, combined with user profiles, to obtain more and broader information related to the current search intent. Deep expanded search results are additional search results related to deeper nodes in the original search result tree obtained through expanded search.

[0098] Specifically, first, based on the user profile information of the current target user, an expanded search operation is performed on the deep-level nodes to obtain deep expanded search results that are semantically related to the deep-level nodes; for example, when the user shows a high degree of attention to the deep-level node of a coffee shop, an expanded search can be performed based on the node and combined with the user profile (such as the user profile shows that the user prefers quiet coffee shops with outdoor seating), and coffee shop information that meets these conditions is retrieved from a wider database, such as quiet coffee shops, coffee shops with outdoor seating, etc. as deep expanded search results.

[0099] Specifically, the correlation between the deep expansion search results and the corresponding deep level nodes is calculated and evaluated to generate a correlation evaluation result, and the deep expansion search results are serialized and sorted based on the correlation evaluation result (sorted from high to low according to the correlation) to improve the relevance and interpretability of the result display.

[0100] Furthermore, the deep expanded search results are updated to the corresponding deep-level nodes in the search result tree based on the serialization results. For example, quiet cafes, coffee shops with outdoor seating, etc. are added as new branches or sub-nodes under the above-mentioned coffee shop nodes and displayed on the display interface (such as displaying these new expanded results in a list form on the HUD).

[0101] The above process can further expand the depth and breadth of search results, provide users with accurate and personalized information, and enable the search result tree to be dynamically updated and expanded. This helps increase the likelihood that users will obtain the information they need, enhancing the practicality of search results and the user experience.

[0102] In summary, the search result display method provided by the present invention has the following technical effects:

[0103] After obtaining the search results, a search result tree containing hierarchical nodes is constructed, and the result tree is displayed in a shallow form on the target display interface of the terminal device, and marks are added to the nodes that can be further expanded; the user's browsing sight behavior data is collected in real time by using the eye tracking device, and the attention heat of each node in the shallow display is calculated based on the data; the nodes are screened and matched based on the attention heat, the search result tree structure and the user portrait to identify the user's real-time search intention; a deep display instruction is generated according to the identified real-time search intention, and the target display interface is driven by the instruction to display the corresponding node in depth, thereby achieving the technical effect of accurate intention recognition and active information push, personalized and efficient search result display.

[0104] Example 2, as Figure 2 This is a schematic diagram of the structure of a search result display system of the present invention. For example, Figure 1 The flowchart of a search result display method of the present invention can be shown as follows: Figure 2 The structure shown is implemented.

[0105] Based on the same concept as the search result display method in the embodiment, the present invention also provides a search result display system including:

[0106] The shallow display module 11 is used to obtain the search results and construct a search result tree including hierarchical nodes, and perform shallow display of the search result tree on the target display interface of the terminal device and perform expandability marking.

[0107] The sight collection and heat calculation module 12 is used to collect the browsing sight behavior data of the target user in real time through the sight tracking device, and calculate the shallow display attention heat set based on the browsing sight behavior data.

[0108] The intention recognition module 13 is used to perform screening and matching based on the attention heat set, the search result tree and the user portrait to identify the real-time search intention of the target user.

[0109] The deep display control module 14 is configured to generate corresponding deep display instructions based on the real-time search intent, and control the target display interface to perform deep display of the search result tree through the deep display instructions.

[0110] In some embodiments, the shallow display module 11 performs the following steps:

[0111] The search result tree generating unit is used to generate a tree display structure according to the association strength and semantic hierarchical relationship of the search results, and output it as the search result tree, wherein the search result tree includes at least one shallow level node and at least one deep level node.

[0112] The expandability flag setting unit is configured to mark the expandability of the shallow-level node under which the deep-level node is subordinate as expandable.

[0113] The shallow level node visualization presentation unit is configured to present the visualization content of the shallow level node on the target display interface according to the expandability markup structure and the relevance between the shallow level node and the search content.

[0114] In some embodiments, the gaze acquisition and heat calculation module 12 performs the following steps:

[0115] A timestamp alignment unit is used to align the timestamps of the target display interface and the gaze tracking device, wherein the gaze tracking device is integrated into the terminal device.

[0116] The eye tracking activation unit is used to synchronously activate the eye tracking device to perform eye tracking on the target user when the shallow presentation starts.

[0117] The browsing sight behavior data generating unit is used to generate the browsing sight behavior data according to the eye tracking results, wherein the browsing sight behavior data includes the sight dwell time, sight trajectory path and number of return glances in each shallow level node area in the target display interface.

[0118] In some embodiments, the gaze acquisition and heat calculation module 12 may further execute the following steps:

[0119] The browsing sight behavior data preprocessing unit is used to preprocess the browsing sight behavior data, and the preprocessing includes normalization processing and dimensionless conversion.

[0120] The confidence analysis and correction weight set generation unit is used to perform confidence analysis on the pre-processed browsing sight behavior data, determine the corresponding confidence factor, and generate a correction weight set.

[0121] The attention heat set generation unit is used to perform weighted calculation on the pre-processed browsing vision behavior data in combination with the modified weight set and the preset data fusion weight parameter set to generate attention heat values ​​corresponding to multiple shallow-level nodes to constitute the attention heat set.

[0122] In some embodiments, the execution steps of the intention recognition module 13 include:

[0123] The user portrait feature extraction unit is used to extract user portrait features from the long-term historical search records of the target user, wherein the user portrait features include keyword preference features and high-frequency topic features.

[0124] The node semantic feature extraction unit is used to traverse the shallow level nodes in the search result tree and extract corresponding node semantic features.

[0125] The feature similarity matching calculation unit is used to traverse the node semantic features based on the user portrait features to perform feature similarity matching calculation and obtain the portrait matching degree.

[0126] The constraint comparison unit is used to compare the portrait matching degree with the portrait matching degree constraint, and to compare the corresponding attention heat value with the attention heat value constraint.

[0127] The real-time retrieval intention determination unit is used to determine the corresponding shallow level node as the real-time retrieval intention if the portrait matching degree satisfies the portrait matching degree constraint and the attention heat value satisfies the attention heat value constraint.

[0128] In some embodiments, the execution steps of the deep display control module 14 include:

[0129] The display layout optimization and folding processing unit is used to optimize the display layout of the target display interface according to the distribution of the attention heat concentration, and fold the shallow level nodes corresponding to low heat values.

[0130] The preloading and prerendering execution unit is used to perform preloading and prerendering on the deep-level nodes under the shallow-level nodes pointed to by the real-time retrieval intention based on the optimized display layout.

[0131] The real-time sight focus detection and floating display unit is used to continuously detect the real-time sight focus of the target user through the sight tracking device, and when the real-time sight focus is focused on any shallow-level node, the content of the corresponding deep-level node is displayed in a floating form.

[0132] In some embodiments, further comprising:

[0133] The deep expanded search unit is used to perform an expanded search based on the user portrait based on deep hierarchical nodes to obtain deep expanded search results.

[0134] The association evaluation and serialization unit is used to evaluate the association between the deep expansion search result and the deep level node, and serialize the deep expansion search result according to the association evaluation result.

[0135] The search result tree updating and displaying unit is used to update the deep expanded search result to the search result tree based on the serialization result and display it.

[0136] It should be understood that the embodiments mentioned in this specification focus on their differences from other embodiments. The specific embodiments in the aforementioned embodiment one are also applicable to a search result display system described in embodiment two. For the sake of brevity of the specification, no further elaboration will be given here.

[0137] It should be understood that the embodiments disclosed in the present invention and the above description can enable those skilled in the art to use the present invention to implement the present invention. At the same time, the present invention is not limited to the embodiments mentioned above. It should be understood that those skilled in the art can still modify the technical solutions described in the above embodiments or replace some of the technical features therein with equivalents; and such modifications or replacements do not deviate from the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention and are all included in the scope of protection of the present invention.

Claims

1. A search result display method, characterized in that: include: Obtaining the search results and constructing a search result tree including hierarchical nodes accordingly, and shallowly displaying the search result tree on the target display interface of the terminal device, and marking it as expandable; Collecting the target user's browsing sight behavior data in real time through an eye tracking device, and calculating the shallow display attention heat set based on the browsing sight behavior data; Perform screening and matching based on the attention heat set, the search result tree and the user portrait to identify the real-time search intention of the target user; A corresponding deep display instruction is generated based on the real-time search intention, and the deep display instruction is used to control the target display interface to perform deep display of the search result tree.

2. A search result display method according to claim 1, characterized in that: Obtaining the search results and constructing a search result tree including hierarchical nodes accordingly, and shallowly displaying the search result tree on the target display interface of the terminal device, and marking it for expandability, including: Generate a tree-like display structure according to the association strength and semantic hierarchical relationship of the search results, and output it as the search result tree, wherein the search result tree includes at least one shallow level node and at least one deep level node; Marking the expandability of the shallow-level node under which the deep-level node is subordinate as expandable; According to the expandability mark structure and the relevance between the shallow level nodes and the search content, the visual content of the shallow level nodes is presented on the target display interface.

3. A search result display method according to claim 2, characterized in that: The target user's browsing behavior data is collected in real time through the gaze tracking device, including: aligning the timestamps of the target display interface and the gaze tracking device, wherein the gaze tracking device is integrated into the terminal device; When the shallow presentation begins, synchronously activating the eye tracking device to perform eye tracking on the target user; The browsing sight behavior data is generated according to the eye tracking result, wherein the browsing sight behavior data includes the sight dwell time, sight trajectory path and number of return glances in each shallow level node area in the target display interface.

4. A search result display method according to claim 3, characterized in that: Calculating a shallow display attention heat set based on the browsing sight behavior data includes: Preprocessing the browsing sight behavior data, wherein the preprocessing includes normalization and dimensionless conversion; Performing confidence analysis on the pre-processed browsing sight behavior data, determining corresponding confidence factors, and generating a modified weight set; The pre-processed browsing sight behavior data is weightedly calculated in combination with the modified weight set and the preset data fusion weight parameter set to generate attention heat values ​​corresponding to multiple shallow-level nodes to constitute the attention heat set.

5. A search result display method according to claim 4, characterized in that: Screening and matching are performed based on the attention heat set, the search result tree, and the user profile to identify the real-time search intention of the target user, including: Extracting user profile features from the target user's long-term historical search records, wherein the user profile features include keyword preference features and high-frequency topic features; Traversing the shallow level nodes in the search result tree and extracting corresponding node semantic features; Based on the user portrait features, traverse the node semantic features to perform feature similarity matching calculation to obtain the portrait matching degree; Comparing the portrait matching degree with the portrait matching degree constraint, and comparing the corresponding attention heat value with the attention heat value constraint; If there is a portrait matching degree that satisfies the portrait matching degree constraint, and the attention heat value satisfies the attention heat value constraint, then the corresponding shallow level node is determined as the real-time retrieval intention.

6. A search result display method according to claim 5, characterized in that: Generating corresponding deep display instructions based on the real-time search intent, and controlling the target display interface to perform deep display of the search result tree through the deep display instructions, including: According to the distribution of the attention heat concentration, the display layout of the target display interface is optimized, and the shallow level nodes corresponding to low heat values ​​are collapsed; Based on the optimized display layout, preloading and prerendering are performed on the deep-level nodes under the shallow-level nodes pointed to by the real-time search intention; The real-time sight focus of the target user is continuously detected by the sight tracking device. When the real-time sight focus is focused on any shallow-level node, the content of the corresponding deep-level node is displayed in a floating form.

7. A search result display method according to claim 1, characterized in that: Also includes: Performing an expanded search based on the user portrait based on deep hierarchical nodes to obtain deep expanded search results; evaluating the relevance between the deep expanded search result and the deep level nodes, and serializing the deep expanded search result according to the relevance evaluation result; The deep expanded search results are updated to the search result tree based on the serialization results and displayed.

8. A search result display system, characterized in that: A method for displaying search results according to any one of claims 1 to 7, comprising: A shallow display module is used to obtain the search results and construct a search result tree including hierarchical nodes, and shallowly display the search result tree on the target display interface of the terminal device and mark it as expandable; The sight collection and heat calculation module is used to collect the target user's browsing sight behavior data in real time through the sight tracking device, and calculate the shallow display attention heat set based on the browsing sight behavior data; An intention recognition module is used to perform screening and matching based on the attention heat set, the search result tree and the user portrait to identify the real-time search intention of the target user; The deep display control module is used to generate corresponding deep display instructions based on the real-time search intention, and control the target display interface to perform deep display of the search result tree through the deep display instructions.

Images