User event indexing method and device, electronic equipment and storage medium

Abstract

The invention relates to a user event indexing method and device, electronic equipment and a storage medium. Receiving retrieval content input by a user through a preset webpage form of a user interface, and obtaining a first index result according to the retrieval content; the method comprises the following steps: acquiring multi-dimensional data of a user in an interaction process by utilizing a sensor, transmitting the multi-dimensional data to a memory, and preprocessing the multi-dimensional data in the memory; the retrieval content and the multi-dimensional data are cleaned and preprocessed, and cache data are obtained; identifying and analyzing the cached data by using a pre-constructed emotion annotation data set to obtain emotion features of the user; the emotion annotation data set is trained through a pre-collected annotation sample set; according to the emotion features, reordering the first index result according to a preset weight to obtain a second index result; and presenting the second index result to the user interface.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a user event indexing method, device, electronic device and storage medium. Background Art

[0002] With the rapid development of computer technology, people can obtain the multimedia information they want by searching the Internet anytime and anywhere. With the increase in the use of search engines, the Internet has become an effective way for users to obtain information. However, the huge amount of data and unreasonable data index structure result in low efficiency of index execution events.

[0003] Based on this technology, there is CN116186333A-user event indexing method, device, electronic device and storage medium. The user event indexing method includes: obtaining event data and at least one user journey of users of an application; based on the event data of the user and the event identifier of the user journey, screening out a target user journey that matches the event data in at least one user journey; based on the event index structure of the target user journey, obtaining the audience conditions of the target user journey; using the audience conditions of the target user journey to perform discrimination processing on the event data to obtain a discrimination result; the scheme executes the execution event of the target user journey based on the discrimination result, thereby improving the efficiency of the user event index execution event; however, the index dimensions of the scheme are traditional dimensions such as time, place, age, etc., and the generated index tags are relatively simple, and personalized recommendations based on user tags are not made. Summary of the invention

[0004] The main purpose of the present invention is to provide a user event indexing method, device, electronic device and storage medium to solve the problem of single index tag in the prior art.

[0005] To achieve the above object, the present invention provides a user event indexing method, comprising the following steps:

[0006] Receiving search content input by a user through a preset webpage form of a user interface, and obtaining a first index result according to the search content;

[0007] Using sensors to obtain multi-dimensional data of the user during the interaction process, transmitting the multi-dimensional data to a memory, and pre-processing the multi-dimensional data in the memory; the multi-dimensional data includes at least one of voice, facial expression and body movement;

[0008] Cleaning and preprocessing the search content and the multi-dimensional data to obtain cache data;

[0009] Using a pre-built emotion annotation dataset, the cached data is identified and analyzed to obtain the user's emotion characteristics; the emotion annotation dataset is trained by a pre-collected annotation sample set;

[0010] Reorder the first index results according to the emotional features according to a preset weight to obtain a second index result;

[0011] The second index result is presented to the user interface.

[0012] Furthermore, while receiving the search content input by the user through the preset webpage form of the user interface, it also includes:

[0013] Use the input event monitoring mechanism to monitor the input status of the preset web page form of the user interface in real time;

[0014] Determine whether to enter user input mode;

[0015] If the user input mode is entered, the sensor is triggered to perform data collection activities;

[0016] Determine whether the user input operation is completed;

[0017] If the user input operation ends, the data collection activity of the sensor is stopped.

[0018] Furthermore, the steps of obtaining multi-dimensional data of the user during the interaction process by using a sensor, transmitting the multi-dimensional data to a memory, and preprocessing the multi-dimensional data in the memory include:

[0019] During the user interaction process, multiple sensors are used to respectively acquire multi-dimensional data with time stamps; the multi-dimensional data includes multiple different types of data, and the data types include at least one of voice data, facial expression data, and body movement data; wherein the voice data is collected by a microphone collection device, the facial expression is collected by a camera device, and the body movement is collected by a camera device,

[0020] Different types of data in the multi-dimensional data are interactively associated with the search content according to timestamps, and the associated multi-dimensional data and the search content are stored in the memory in a time sequence.

[0021] Furthermore, the steps of identifying and analyzing the cached data using a pre-built emotion annotation dataset to obtain the user's emotion characteristics include:

[0022] Matching each type of data in the cached data with the emotional features in the pre-built emotional annotation data set one by one, and outputting a single emotional feature matching value;

[0023] All types of data in the cached data are calibrated with emotional features using a pre-constructed emotional annotation dataset, and an overall emotional feature matching value is output;

[0024] According to each of the individual emotion feature matching values ​​and the overall emotion feature matching value, the current accurate emotion feature of the cached data is obtained and then marked;

[0025] The emotion annotation dataset includes emotion tags and text datasets, which are divided into three modules: text, speech and image.

[0026] Furthermore, the step of reordering the first index results according to the preset weights based on the emotional features to obtain the second index results includes:

[0027] Sort each search result by position according to the frequency of occurrence of the search content in the first index result;

[0028] Adjusting the order of the first index results according to the second preset rule to obtain a second index result;

[0029] Presenting the second index result to the user interface;

[0030] The second preset rule is to calculate the sentiment feature of each search result in the first index result, and assign a sentiment intensity score to each sentiment feature in the first index result and the sentiment feature of the cached data through the sentiment annotation data set;

[0031] Obtaining the sentiment score of the first index result and the sentiment score of the cached data;

[0032] The cached data sentiment score and the first index result sentiment score are sorted according to the matching degree to obtain a second index result; and the second index result is presented to a user interface.

[0033] Among them, the rules for presetting weights are:

[0034] First, the TF-IDF value is calculated: the TF-IDF value of each sentiment feature in the first index result is calculated according to the TF-IDF algorithm.

[0035] TF: The number of times a sentiment feature appears in a specific index result.

[0036] IDF: The logarithm of the inverse number of documents in which a sentiment feature appears in the entire dataset or in a specific domain.

[0037] The calculation formula is: TF-IDF = term frequency (TF) * inverse document frequency (IDF)

[0038] Then, assign sentiment intensity scores: Assign a sentiment intensity score to each sentiment feature. The score is the output based on the lookup results of the sentiment annotation dataset. Use a pre-built sentiment annotation dataset, where each sentiment word or phrase is assigned a sentiment intensity value, for example, +1 to +3 for positive sentiment intensity, -1 to -3 for negative sentiment intensity, and 0 for neutral. By looking up the sentiment annotation dataset, quickly assign sentiment intensity scores to the sentiment features in the document.

[0039] Finally, comparative sorting is performed: sorting is performed according to the matching degree between the sentiment score of the cached data and the sentiment score of the first index result.

[0040] Furthermore, the step of presenting the second index result to the user interface further includes:

[0041] Establish a real-time user feedback mechanism to obtain user feedback satisfaction with the second index results;

[0042] Capture user network behavior path information to obtain user portrait data; the user portrait data includes at least one of personal information, historical behavior, browsing preferences, cognitive style and knowledge background;

[0043] The second index result is sorted again according to the user portrait data to obtain a third index result, the third index result is used as the final index result, and the index result displayed on the user interface is updated to the third index result.

[0044] The present invention also provides a user event indexing device, comprising:

[0045] A receiving unit receives search content input by a user through a preset web page form of a user interface, and obtains a first index result according to the search content;

[0046] an acquisition unit, which uses a sensor to acquire multi-dimensional data of the user during the interaction process, transmits the multi-dimensional data to a memory, and performs dimension classification processing on the data in the memory;

[0047] A data processing unit cleans and preprocesses the search content and the multi-dimensional data to obtain cache data;

[0048] A labeling unit, using a pre-built emotion labeling data set, identifies and analyzes the cached data to obtain the user's emotion characteristics;

[0049] A retrieval unit, reordering the first index results according to a preset weight based on the sentiment feature to obtain a second index result;

[0050] The display unit presents the second index result on the user interface.

[0051] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned user event indexing method when executing the computer program.

[0052] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned user event indexing method are implemented.

[0053] The user event indexing method, device, electronic device and storage medium provided by the present invention have the following beneficial effects:

[0054] (1) By introducing new dimensions such as user behavior patterns and emotional characteristics for indexing, we can establish richer index tags compared to traditional indexing methods based only on dimensions such as time and location, thereby significantly improving the accuracy and efficiency of indexing. This helps to more accurately match user needs and provide relevant information.

[0055] (2) By analyzing the emotional characteristics of users using the emotion annotation dataset and combining it with user portrait data and real-time feedback mechanism, personalized index result sorting can be achieved. Users will be more likely to obtain information that matches their interests, preferences, and emotional states, thus improving the user experience.

[0056] (3) Establish a real-time user feedback mechanism to allow users to evaluate and provide feedback on index results. Based on user feedback and user profile data, the ranking weight of index results can be dynamically adjusted to ensure that the index results are always consistent with user needs and preferences. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 is a flowchart of a user event indexing method in one embodiment of the present invention;

[0058] Figure 2 is a structural block diagram of a user event indexing device in one embodiment of the present invention;

[0059] Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0060] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0062] Reference Figure 1 , is a flow chart of a user event indexing method proposed by the present invention, comprising the following steps:

[0063] S1, receiving search content input by a user through a preset web page form of a user interface, and obtaining a first index result according to the search content;

[0064] S2, using a sensor to obtain multi-dimensional data of the user during the interaction process, transmitting the multi-dimensional data to a memory, and pre-processing the multi-dimensional data in the memory; the multi-dimensional data includes at least one of voice, facial expression and body movement;

[0065] S3, cleaning and preprocessing the search content and the multi-dimensional data to obtain cache data;

[0066] S4, using a pre-constructed emotion annotation dataset to identify and analyze the cached data to obtain the user's emotion characteristics; the emotion annotation dataset is trained by a pre-collected annotation sample set;

[0067] S5, re-sorting the first index results according to the preset weights based on the emotional features to obtain second index results;

[0068] S6: Present the second index result to the user interface.

[0069] For S1, specifically:

[0070] The step of receiving search content input by a user through a preset webpage form of a user interface and obtaining a first index result according to the search content includes:

[0071] The system receives search query content entered by the user through a predefined web form in the user interface, submits the query content to the back-end processing system, and uses a pre-established indexing mechanism to retrieve a set of records most relevant to the query content from a huge data set. The record set is the first index result obtained based on the user's search content.

[0072] While implementing S1, it also includes:

[0073] Use the input event monitoring mechanism to monitor the input status of the preset web page form of the user interface in real time;

[0074] Determine whether to enter user input mode;

[0075] If the user input mode is entered, the sensor is triggered to perform data collection activities;

[0076] Determine whether the user input operation is completed;

[0077] If the user input operation ends, the data collection activity of the sensor is stopped.

[0078] In the specific implementation: the mouse clicks on the search box as the key point of event monitoring for judging whether to enter the user input mode, and the listening function of the input event is bound through JavaScript. Once the mouse clicks on the search box, the input event is triggered, it is judged that the user input mode has been entered, and the sensor is started to perform data collection activities; the user interface sets an index button or one of the specific keys on the keyboard as the key point of event monitoring for judging whether the user input operation is completed. Once the mouse clicks on the index button or one of the specific keys on the keyboard, the input event is triggered, it is judged that the user input operation is completed, and the data collection activity of the sensor started before is stopped.

[0079] For S2, specifically:

[0080] The steps of using a sensor to obtain multi-dimensional data of a user during an interaction process, transmitting the multi-dimensional data to a memory, and preprocessing the data in the memory include:

[0081] During the user interaction process, multiple sensors are used to respectively acquire multi-dimensional data with time stamps; the multi-dimensional data includes multiple different types of data, and the data types include at least one of voice data, facial expression data, and body movement data; wherein the voice data is collected by a microphone collection device, the facial expression is collected by a camera device, and the body movement is collected by a camera device,

[0082] Different types of data in the multi-dimensional data are interactively associated with the search content according to timestamps, and the associated multi-dimensional data and the search content are stored in the memory in a time sequence.

[0083] In the specific implementation: after the sensor is triggered, the system starts to automatically perform data collection operations to obtain synchronous data of the specified type with a timestamp; collects voice data through the built-in micro microphone of the device to capture sound signals in the environment, especially human voice; the camera collects image data of facial expressions and body movements, and captures video images, especially human facial expressions and body movements; the voice data and image data collected by the sensor are interactively processed with the search content based on timestamps, and are transmitted to the storage in different data types, such as voice and image.

[0084] For S3, specifically:

[0085] The steps of cleaning and preprocessing the multi-dimensional data to obtain cached data include:

[0086] For text data, after performing cleaning operations such as removing invalid data, removing stop words, algorithmically correcting spelling mistakes, and desensitizing sensitive information, and then performing preprocessing tasks such as word segmentation, part-of-speech tagging, and named entity recognition, it is cached in the memory;

[0087] For speech data, after performing cleaning operations such as removing background noise and silence and segmenting continuous speech into independent segments, and then performing preprocessing tasks such as enhancing the high-frequency part of the speech signal and extracting useful features from the speech signal using Mel-frequency cepstral coefficients, it is cached in the memory;

[0088] For image data, after deleting duplicate images, filtering low-quality, non-standard, and non-compliant images, and then performing preprocessing tasks such as standardizing the image format and geometric operations to adjust the image size and orientation, it is cached in the memory.

[0089] In specific implementation: The cleaning operations for text data include removing invalid data, removing stop words, algorithmically correcting spelling mistakes, and desensitizing sensitive information; Removing invalid data, deleting invalid information such as irrelevant characters, garbled characters, and special symbols in the text; Removing stop words, removing words that frequently appear in the text but are not substantially helpful for understanding the text content, such as "de", "le", "zai"; Algorithmically correcting spelling mistakes, automatically detecting and correcting spelling mistakes in the text using natural language processing algorithms; Desensitizing sensitive information, encrypting or replacing sensitive information in the text, such as personal identity information, privacy data, and phone numbers, to protect user privacy; The preprocessing tasks for text data include word segmentation, part-of-speech tagging, and named entity recognition; Word segmentation, splitting the text into meaningful lexical units for subsequent part-of-speech tagging and named entity recognition; Part-of-speech tagging, tagging the part of speech for each lexical unit to reflect the grammatical function of the word in the text, such as nouns, verbs, adjectives; Named entity recognition, identifying named entities in the text, such as person names, place names, and organization names.

[0090] The cleaning operations for speech data include removing background noise, silence and independently segmenting continuous speech; removing background noise and silence, reducing or eliminating the background noise and silence parts in the speech signal through signal processing algorithms to improve speech quality; independently segmenting continuous speech, dividing the continuous speech signal into independent speech segments; the preprocessing of speech data includes enhancing the high-frequency part of the speech signal and extracting useful features from the speech signal using Mel-frequency cepstral coefficients (MFCC); enhancing the high-frequency part of the speech signal, enhancing the high-frequency part of the speech signal through signal processing algorithms to improve the clarity and intelligibility of the speech.

[0091] Cleaning operations for image data include deleting duplicate images, filtering low-quality, non-standard, and non-compliant images; deleting duplicate images, removing repeated images in the data set to reduce data redundancy; filtering low-quality, non-standard, and non-compliant images, such as blurred, ghosting, and overexposed images. Preprocessing of image data includes format conversion, resizing, normalization, and data enhancement; format conversion, converting images to a unified format, such as JPEG, PNG; resizing, adjusting the size of the image to adapt to different application scenarios or model input requirements; normalization, normalizing or standardizing the pixel values ​​of the image to improve the training efficiency and performance of the model; data enhancement, increasing the diversity of images through operations such as rotation, translation, scaling, and flipping, and improving the generalization ability of the model.

[0092] For S4, specifically:

[0093] The steps of identifying and analyzing the cached data using a pre-built emotion annotation dataset to obtain the user's emotion characteristics include:

[0094] Matching each type of data in the cached data with the emotional features in the pre-built emotional annotation data set one by one, and outputting a single emotional feature matching value;

[0095] All types of data in the cached data are calibrated with emotional features using a pre-constructed emotional annotation dataset, and an overall emotional feature matching value is output;

[0096] According to the matching values ​​of the individual emotional features and the overall emotional feature matching values, the current accurate emotional features of the cached data are obtained and then labeled; the emotional labeling data set includes emotional tags and text data sets, which are divided into three modules: text, voice and image.

[0097] In specific implementation: The sentiment labeling dataset is a collection of sentiment labels and corresponding data. The dataset is manually labeled, and corresponding sentiment labels are assigned to each data item, such as positive, negative, and neutral. The dataset is divided into three modules: text, speech, and image, corresponding to different data types and corresponding sentiment recognition tasks; the multi-dimensional data in the cache, such as text, speech, and image data, are matched with the data in the sentiment labeling dataset. This involves the feature extraction process, that is, converting the original data into a format that the model can understand, such as converting text into word vectors, and converting speech and images into feature vectors; in text processing, the recurrent neural network (RNN) analyzes the sequential relationship between words and understands the overall meaning of the sentence. In speech processing, the recurrent neural network (RNN) processes the time series characteristics of speech signals and identifies emotional fluctuations in speech. In image processing, a recurrent neural network (RNN) processes sequence data (such as video frames) or sequences of image features in an image; for data from different dimensions (text, voice, image), the model performs emotion recognition separately and extracts their respective emotion features, synthesizes these features, and uses a weighted average algorithm to derive an accurate representation of the current user's emotion, assigning one or more emotion labels to the user's current emotional state, where the emotion labels are based on predefined emotion categories, such as happiness, sadness, and anger.

[0098] For S5, specifically:

[0099] The step of reordering the first index results according to the preset weights based on the emotional features to obtain the second index results includes:

[0100] Indexing is performed according to user search content, and the obtained index results are preliminarily sorted according to the strength of relevance to the index content. Based on the preliminary sorting, the final index results are presented after further adjustment according to the emotional tendency of the emotional features.

[0101] The rules for the preset weights are:

[0102] First, calculate the TF-IDF value: Calculate the TF-IDF value of each sentiment feature in the document according to the TF-IDF algorithm.

[0103] The calculation formula is: TF-IDF = term frequency (TF) * inverse document frequency (IDF)

[0104] TF: The number of times a sentiment feature appears in a specific index result.

[0105] IDF: The logarithm of the inverse number of documents in which a sentiment feature appears in the entire dataset or in a specific domain.

[0106] Then, assign sentiment intensity scores: assign a sentiment intensity score to each sentiment feature, which is the output of the search results based on the sentiment annotation dataset. Use a pre-built sentiment annotation dataset, where each sentiment word or phrase is assigned a sentiment intensity value, for example, +1 to +3 represents positive sentiment intensity, -1 to -3 represents negative sentiment intensity, and 0 represents neutral; by searching the sentiment annotation dataset, quickly assign sentiment intensity scores to sentiment features in the document.

[0107] Finally, a comparative ranking is performed: considering the importance of sentiment features in the document and the strength of sentiment tendency, the TF-IDF value of each sentiment feature is multiplied by its corresponding sentiment intensity score, and the results of all sentiment features are summed or averaged.

[0108] For S6, specifically:

[0109] The step of presenting the second index result to the user interface is:

[0110] The second index result is formatted into a form suitable for user interface display, key information of the result is extracted, such as at least one of the title, summary, and link, and a picture or video thumbnail can be attached if necessary, and then rendered to the corresponding area of ​​the user interface.

[0111] In specific implementation:

[0112] The second index result is formatted into a form suitable for user interface (UI) display, such as adjusting at least one of the structure, type, and format of the data to match the expected input of the UI component; key information is extracted from each search result, wherein the key information is at least one of the title, summary, and link. In addition to the above key information, other information can also be extracted according to specific needs, such as thumbnails of pictures and videos, authors, release times, etc.; the extracted key information is rendered to the corresponding area of ​​the user interface through JavaScript language, and finally presented to the user interface.

[0113] For S7 and later versions, the details are as follows:

[0114] S8, establishing a user real-time feedback mechanism to obtain user feedback satisfaction with the second index result;

[0115] S9, capturing the user's network behavior path information and obtaining user portrait data; the user portrait data includes at least one of personal information, historical behavior, browsing preference, cognitive style and knowledge background;

[0116] S10, re-sorting the second index result according to the user portrait data to obtain a third index result, taking the third index result as the final index result, and updating the index result displayed on the user interface to the third index result.

[0117] For S8, specifically:

[0118] The steps to capture user network behavior path information and obtain user portrait data include:

[0119] The user portrait data includes at least one of personal information, historical behavior, browsing preference, cognitive style and knowledge background;

[0120] Use crawler technology to capture user behavior data from the Internet platform, including search behavior, purchase behavior, and usage behavior, and call the API interface of the target website to obtain stable data; obtain basic data on user background through social media platforms; obtain psychological characteristics of user interests, preferences, and emotions through social media interactions and evaluations; after removing invalid and duplicate data, integrate user data from different data sources to obtain user behavior data sets and build user portrait data.

[0121] In specific implementation: use automated scripts to capture user behavior data from Internet platforms. For example, users’ search behavior on search engines, purchase behavior on e-commerce platforms, and usage behavior on various applications or websites. Crawler technology collects a large amount of user data by simulating user browsing behavior, and obtains stable and structured data by calling the API interface of the target website; obtains basic user data such as age, gender, geographic location, and occupation through social media platforms; analyzes user interactive behaviors on social media, such as likes, comments, and sharing, to obtain user interests and preferences; collects user evaluation information on products, services, and content to further understand user preferences and emotional tendencies.

[0122] After collecting a large amount of user data, data cleaning is performed to remove invalid and duplicate data, and user data from different data sources are integrated to form a complete user behavior data set. Based on the user behavior data set, cluster analysis is used to construct user portrait data.

[0123] For S9, specifically:

[0124] The step of establishing a user real-time feedback mechanism to provide feedback on the index results includes:

[0125] An easily accessible scoring system real-time feedback portal is set in the preset web page form of the user interface to obtain user feedback satisfaction with the second index result.

[0126] In specific implementation:

[0127] On the page showing the index results, a real-time feedback portal that is easy to access and operate is designed. This portal is a system form of star rating and text feedback. It guides users to participate in feedback through prompt animations and collects user feedback data in real time. When the user clicks the feedback option, the system immediately records the user's feedback result and the corresponding index result identifier, and stores the user's feedback data in the database. The collected user feedback data is statistically analyzed to identify the user's satisfaction with the second index result. For example, the satisfaction rating ratio of each index result is calculated. If the feedback data contains text comments, sentiment analysis can also be performed to gain a deeper understanding of the user's emotional tendencies and specific opinions.

[0128] For S10, specifically:

[0129] The steps of re-sorting the second index result according to the user portrait data to obtain a third index result, using the third index result as the final index result, and updating the index result displayed on the user interface to the third index result include:

[0130] An index sorting system is constructed by assigning weights to the user portrait data based on a first preset rule; the first preset rule is that the weights decrease in the order of historical behavior data, browsing preferences, knowledge background, cognitive style, and personal information.

[0131] In specific implementation: When building a search ranking system, the user portrait data is systematically weighted, specifically: first, the user's historical behavior data accounts for 35%, which directly and significantly reflects the user's interest trends and demand patterns, and is given the highest weight; second, browsing preferences account for 25%, which indirectly reveals the user's deep interests through the user's specific behaviors during the browsing process (such as dwell time, interaction frequency, etc.), and its weight is second; third, knowledge background accounts for 20%. Although this dimension does not directly determine the search intent, it plays a key role in understanding the deep content, professional terms and domain knowledge that the user may be interested in, so it also occupies a certain weight; third, cognitive style accounts for 15%, which is related to the user's information processing preferences and reading habits, and has guiding significance for optimizing the presentation of search results, but compared with the first three, its weight is lower; finally, personal information accounts for 5% as a supplementary dimension, which plays a role in specific scenarios (such as regional recommendations), and its weight allocation is the lowest. The weight allocation of user portrait data follows the logical order from direct relevance to indirect influence, from core interests to auxiliary information.

[0132] For example, for a certain user, if his historical behavior shows that he has a high interest in technology content, and real-time feedback also shows that he is highly satisfied with the technology index results, then the ranking weight of technology content in the index results can be increased accordingly.

[0133] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0134] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0135] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0136] In summary, the search content input by the user is received through the preset web page form of the user interface, and the first index result is obtained according to the search content; the multi-dimensional data of the user in the interaction process is obtained by using a sensor, the multi-dimensional data is transmitted to the memory, and the data in the memory is dimensionally classified; the search content and the multi-dimensional data are cleaned and preprocessed to obtain cache data; the cache data is identified and analyzed by using a pre-constructed emotion annotation data set to obtain the user's emotional characteristics; the first index result is re-sorted according to the preset weight according to the emotional characteristics to obtain the second index result; the display unit presents the second index result to the user interface; a user real-time feedback mechanism is established to provide feedback on the index result; the user's network behavior path information is captured to obtain user portrait data; the second index result is re-sorted according to the user portrait data to obtain a third index result, the third index result is used as the final index result, and the index result displayed on the user interface is updated to the third index result, so as to realize personalized recommendations based on user tags, introduce new dimensions of user behavior patterns and emotional characteristics for indexing, establish richer index tags, and improve the accuracy and efficiency of indexing.

[0137] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0138] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0139] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A user event indexing method, characterized in that: The following steps are involved: Receiving search content input by a user through a preset webpage form of a user interface, and obtaining a first index result according to the search content; Using sensors to obtain multi-dimensional data of the user during the interaction process, transmitting the multi-dimensional data to a memory, and pre-processing the multi-dimensional data in the memory; The multi-dimensional data includes at least one of voice, facial expression and body movement; Cleaning and preprocessing the search content and the multi-dimensional data to obtain cache data; Using a pre-built emotion annotation dataset, the cached data is identified and analyzed to obtain the user's emotion characteristics; The emotion annotation dataset is trained by using a pre-collected annotation sample set; Reorder the first index results according to the emotional features according to a preset weight to obtain a second index result; The second index result is presented to the user interface.

2. The user event indexing method according to claim 1, characterized in that: The steps after presenting the second index result to the user interface include: Establish a real-time user feedback mechanism to obtain user feedback satisfaction with the second index results; Capture user network behavior path information to obtain user portrait data; the user portrait data includes at least one of personal information, historical behavior, browsing preferences, cognitive style and knowledge background; The second index result is sorted again according to the user portrait data to obtain a third index result, the third index result is used as the final index result, and the index result displayed on the user interface is updated to the third index result.

3. The user event indexing method according to claim 2, characterized in that: The step of re-sorting the second index results according to the user portrait data to obtain a third index result includes: An index sorting system is constructed by assigning weights to the user portrait data based on a first preset rule; the first preset rule is that the weights decrease in the order of historical behavior data, browsing preferences, knowledge background, cognitive style, and personal information.

4. The user event indexing method according to claim 1, characterized in that: While receiving the search content input by the user through the preset web page form of the user interface, it also includes: Use the input event monitoring mechanism to monitor the input status of the preset web page form of the user interface in real time; Determine whether to enter user input mode; If the user input mode is entered, the sensor is triggered to perform data collection activities; Determine whether the user input operation is completed; If the user input operation ends, the data collection activity of the sensor is stopped.

5. The user event indexing method according to claim 1, characterized in that: The steps of using a sensor to obtain multi-dimensional data of a user during an interaction process, transmitting the multi-dimensional data to a memory, and preprocessing the multi-dimensional data in the memory include: During the user interaction process, multiple sensors are used to respectively acquire multi-dimensional data with time stamps; the multi-dimensional data includes multiple different types of data, and the data types include at least one of voice data, facial expression data, and body movement data; wherein the voice data is acquired by a microphone acquisition device, the facial expression is acquired by a camera device, and the body movement is acquired by a camera device, Different types of data in the multi-dimensional data are interactively associated with the search content according to timestamps, and the associated multi-dimensional data and the search content are stored in the memory in a time sequence.

6. The user event indexing method according to claim 1, characterized in that: The steps of identifying and analyzing the cached data using a pre-built emotion annotation dataset to obtain the emotion characteristics of the user include: Matching each type of data in the cached data with the emotional features in the pre-built emotional annotation data set one by one, and outputting a single emotional feature matching value; All types of data in the cached data are calibrated with emotional features using a pre-constructed emotional annotation dataset, and an overall emotional feature matching value is output; According to each of the individual emotion feature matching values ​​and the overall emotion feature matching value, the current accurate emotion feature of the cached data is obtained and then marked; The emotion annotation dataset includes emotion tags and text datasets, which are divided into three modules: text, speech and image.

7. The user event indexing method according to claim 1, characterized in that: The step of reordering the first index results according to the preset weights based on the emotional features to obtain the second index results includes: Sort each search result by position according to the frequency of occurrence of the search content in the first index result; Adjusting the order of the first index results according to the second preset rule to obtain a second index result; Presenting the second index result to the user interface; The second preset rule is to calculate the sentiment feature of each search result in the first index result, and assign a sentiment intensity score to each sentiment feature in the first index result and the sentiment feature of the cached data through the sentiment annotation data set; Obtaining the sentiment score of the first index result and the sentiment score of the cached data; The cached data sentiment score and the first index result sentiment score are sorted according to the matching degree to obtain a second index result; and the second index result is presented to a user interface.

8. A user event indexing device, characterized in that: include: A receiving unit receives search content input by a user through a preset web page form of a user interface, and obtains a first index result according to the search content; an acquisition unit, which uses a sensor to acquire multi-dimensional data of the user during the interaction process, transmits the multi-dimensional data to a memory, and performs dimension classification processing on the data in the memory; A data processing unit cleans and preprocesses the search content and the multi-dimensional data to obtain cache data; A labeling unit, using a pre-built emotion labeling data set, identifies and analyzes the cached data to obtain the emotion characteristics of the user; A retrieval unit, reordering the first index results according to a preset weight based on the sentiment feature to obtain a second index result; The display unit presents the second index result on the user interface.

9. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the user event indexing method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the user event indexing method according to any one of claims 1 to 7 are implemented.

Images