A media information acquisition method and system, an electronic device, and a storage medium

Abstract

The present disclosure provides a media information acquisition method and system, electronic equipment and storage medium, relates to the technical field of information processing, and aims to solve the timeliness and comprehensiveness problems faced by news reporters when searching news across platforms. The method uses crawler technology to automatically crawl news text data from multiple data source platforms to form a data set, and classifies and quality evaluates the data through a label classification model and a scoring system. At the same time, a user portrait is constructed to accurately reflect user preferences and needs, and then target data meeting the user portrait is matched and pushed according to the type label and quality score of the data. This technical solution not only significantly improves the timeliness and comprehensiveness of news media information acquisition, ensuring that news reporters can quickly grasp the latest information, but also meets the specific needs of news reporters through personalized pushing, reduces the search difficulty, and improves the work efficiency.

Description

Technical Field

[0001] This disclosure relates to the field of information processing technology, and in particular to a method, system, electronic device and storage medium for acquiring media information. Background Technology

[0002] Journalists face significant challenges in finding story topics, needing to sift through numerous platforms (such as public accounts, government websites, financial news portals, and corporate websites) to find valuable information. However, the news industry demands high timeliness, and cross-platform searches lead to information delays, impacting the timeliness of reporting. Furthermore, the wide and complex range of information sources makes it easy to overlook important information through manual searches, affecting the comprehensiveness and accuracy of reports. In addition, each journalist has specific news needs, but existing search methods do not support personalized customization, limiting the selection of news leads. Cross-platform search skills pose a challenge for newcomers; a lack of guidance and tools increases learning costs and reduces work efficiency. Summary of the Invention

[0003] This disclosure provides a method, system, electronic device, and storage medium for acquiring media information, in order to at least solve the above-mentioned technical problems existing in the prior art.

[0004] According to a first aspect of this disclosure, a method for obtaining media information is provided, the method comprising:

[0005] Data was scraped from multiple data source platforms to construct a dataset; the data consisted of news text data.

[0006] The data in the dataset are classified using a label classification model to obtain at least one type label for each data point.

[0007] The data is evaluated using a scoring system to obtain a quality score.

[0008] Obtain user information and build user profiles;

[0009] Based on the type label and quality score of the data, at least one target data that matches the user profile is matched from the dataset and feedback is provided.

[0010] In one possible implementation, the step of scraping data from multiple data source platforms and constructing a dataset includes:

[0011] Web scraping technology was used to extract source data from government websites, financial media, company websites, WeChat official accounts, and social media platforms.

[0012] The source data is preprocessed to obtain the data, and the data is distributed and stored using a big data processing framework and a distributed storage system to construct the dataset.

[0013] In one possible implementation, classifying the data in the dataset using a label classification model to obtain at least one type label corresponding to each data point includes:

[0014] Obtain a training dataset, which includes news text data to be trained;

[0015] Based on a predefined set of multi-dimensional type labels, match at least one dimension of type label to the news text data to be trained;

[0016] Extract keywords and themes from the news text data to be trained to obtain content features;

[0017] Based on the content features and type labels of the news text data to be trained, the initial classification model is trained to obtain the label classification model;

[0018] Using the label classification model, at least one type label corresponding to the news text data in the dataset is obtained.

[0019] In one possible implementation, the step of using a scoring system to assess the quality of the data and obtain a quality score for the data includes:

[0020] The data is scored across multiple dimensions, with corresponding weights assigned to each dimension; the scoring dimensions include content score and influence score.

[0021] The quality score of the data is calculated based on the score and weight of each dimension of the data.

[0022] In one possible implementation, the content rating includes evaluating the accuracy, depth, coherence, and originality of the data, including:

[0023] The accuracy assessment of the data includes: obtaining at least one trusted data source, integrating the text content in the trusted data source that matches the data, performing a logical consistency check on the data and the integrated text content, and obtaining a score for the accuracy of the data.

[0024] The in-depth evaluation of the data includes: calculating the information density and the range of topics covered by the data, and determining a depth score for the data;

[0025] The evaluation of the coherence of the data includes: training a deep learning network including an edge loss function and negative sampling using continuous sentence pairs and incoherent sentence pairs, and using the deep learning network to obtain a score of the coherence of the data;

[0026] The evaluation of the originality of the data includes: using a trained Word2Vec model to convert the text of each data in the dataset into word vectors and aggregating them to obtain article vectors of the data; calculating the similarity between two data based on the article vectors; and determining the score of the originality of the data based on the similarity.

[0027] In one possible implementation, the influence score includes: importance and real-time dissemination, including:

[0028] The assessment of the importance of the data includes: obtaining the number of citations and media citations of the data, and determining the importance score of the data based on the number of citations and media citations;

[0029] Predicting the real-time dissemination of the data includes: obtaining the interaction index of the data within a preset time period, and predicting the score of the real-time dissemination of the data based on the interaction index.

[0030] In one possible implementation, obtaining user information and constructing a user profile includes:

[0031] The user information is obtained by acquiring the user's search history, behavioral characteristics, and points of interest;

[0032] Based on the user information, the user's preference information and behavioral patterns are determined to obtain the user profile.

[0033] According to a second aspect of this disclosure, a media information acquisition system is provided, the system comprising:

[0034] The data crawling module is used to crawl data from multiple data source platforms and build a dataset; the data is news text data.

[0035] The classification module is used to classify the data in the dataset using a label classification model and obtain at least one type label for each data point.

[0036] The scoring module is used to evaluate the quality of the data using a scoring system and obtain a quality score for the data.

[0037] The user profile building module is used to obtain user information and build user profiles.

[0038] The feedback module is used to match at least one target data that matches the user profile from the dataset based on the data's type label and quality score, and then provide feedback.

[0039] According to a third aspect of this disclosure, an electronic device is provided, comprising:

[0040] At least one processor; and

[0041] A memory communicatively connected to the at least one processor; wherein,

[0042] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the methods described in this disclosure.

[0043] According to a fourth aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions for causing the computer to perform the methods described in this disclosure.

[0044] This disclosure discloses a media information acquisition method, system, electronic device, and storage medium that automatically crawls news text data from multiple data source platforms using web crawling technology to construct a dataset. Next, a label classification model is used to classify the data in the dataset, assigning at least one type label to each data point. Then, a scoring system is employed to evaluate the quality of the data, calculating a quality score for each news text data point. Simultaneously, user information is collected and analyzed to construct user profiles, accurately reflecting user preferences and needs. Finally, based on the data's type labels and quality scores, target data matching the user profile is matched from the dataset and fed back to the user through display devices or voice broadcasts. This disclosed technical solution significantly improves the timeliness, comprehensiveness, and accuracy of news media information acquisition through automatic crawling, classification evaluation, and personalized push notifications, ensuring that journalists can quickly grasp the latest information and avoid information delays. At the same time, the accurate classification and quality scoring mechanism guarantees comprehensive information coverage and high reliability. Furthermore, personalized push notifications based on user profiles meet the specific needs of journalists, reduce search difficulty, and improve work efficiency. In summary, this solution brings a more efficient and accurate way to acquire media information to the news industry.

[0045] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0046] The above and other objects, features, and advantages of this disclosure will become readily apparent from the following detailed description of exemplary embodiments, taken in conjunction with the accompanying drawings. Several embodiments of this disclosure are illustrated in the drawings by way of example and not limitation, in which:

[0047] In the accompanying drawings, the same or corresponding reference numerals indicate the same or corresponding parts.

[0048] Figure 1 This illustration shows the implementation flow of the media information acquisition method according to an embodiment of the present disclosure. Figure 1 ;

[0049] Figure 2 This illustration shows the implementation flow of the media information acquisition method according to an embodiment of the present disclosure. Figure 2 ;

[0050] Figure 3 This illustration shows the implementation flow of the media information acquisition method according to an embodiment of the present disclosure. Figure 3 ;

[0051] Figure 4 This illustration shows the implementation flow of the media information acquisition method according to an embodiment of the present disclosure. Figure 4 ;

[0052] Figure 5 This illustration shows the implementation flow of the media information acquisition method according to an embodiment of the present disclosure. Figure 5 ;

[0053] Figure 6 A schematic diagram of the structure of the media information acquisition system according to an embodiment of this disclosure is shown;

[0054] Figure 7 A schematic diagram of the composition structure of an electronic device according to an embodiment of the present disclosure is shown. Detailed Implementation

[0055] To make the objectives, features, and advantages of this disclosure more apparent and understandable, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without creative effort are within the scope of protection of this disclosure.

[0056] This disclosure provides a method for obtaining media information, such as Figure 1 As shown, the method includes:

[0057] Step 101: Extract data from multiple data source platforms and construct a dataset; the data consists of news text data.

[0058] In this example, web scraping technology is used to automatically extract news text data from multiple data source platforms. The extracted data is then integrated into a dataset for use in subsequent steps.

[0059] Step 102: Use a label classification model to classify the data in the dataset and obtain at least one type label for each data point.

[0060] In this example, a label classification model is used to classify news text data in the dataset. Based on natural language processing techniques and machine learning algorithms, the label classification model can identify feature information in news text data and classify the news text data into corresponding labels according to the feature information. Each news text data will be assigned at least one type label (such as content-based labels, sentiment analysis-based labels, and event type-based labels, etc.).

[0061] Step 103: Use a scoring system to assess the quality of the data and obtain a quality score for the data.

[0062] In this example, to improve the quality of the news feed, a scoring system is used to evaluate the quality of each news text data, comprehensively considering multiple aspects such as the news's authenticity, accuracy, timeliness, and newsworthiness. Through calculation, each news text data receives a quality score, which reflects the overall quality and value of the news text data.

[0063] Step 104: Obtain user information and build user profiles.

[0064] In this example, user profiles are constructed by collecting and analyzing information such as users' search history, behavioral characteristics, and interests. User profiles are an accurate reflection of user preferences and needs, which helps to deliver news based on users' personalized needs.

[0065] Step 105: Based on the data type label and quality score, match at least one target data that matches the user profile from the dataset and provide feedback.

[0066] In this example, firstly, based on the user's preferences in the user profile, multiple data points with corresponding category tags are matched from the dataset. Then, the data points within each category tag are sorted according to their quality scores. Finally, the target data whose quality scores match the user's needs is fed back to the user, for example, by displaying the target data on the user's display device or by reading the target data aloud. This target data consists of high-quality news that has been collected and filtered and deemed to be of the user's greatest interest, and is therefore pushed to the user for feedback. In this way, users can receive news pushes that match their preferences and needs, thereby improving information search efficiency.

[0067] This disclosure provides a method for acquiring media information. It utilizes web crawling technology to automatically extract news text data from multiple data source platforms, constructing a dataset. Next, a label classification model is used to categorize the data in the dataset, assigning at least one type label to each data point. Then, a scoring system is employed to evaluate the quality of the data, calculating a quality score for each news text data point. Simultaneously, user information is collected and analyzed to construct user profiles, accurately reflecting user preferences and needs. Finally, based on the data's type labels and quality scores, target data matching the user profile is matched from the dataset and delivered to the user via display device or voice broadcast. This disclosed technical solution significantly improves the timeliness, comprehensiveness, and accuracy of news media information acquisition through automatic crawling, classification evaluation, and personalized push notifications, ensuring that journalists can quickly grasp the latest information and avoid information delays. Furthermore, the precise classification and quality scoring mechanism guarantees comprehensive information coverage and high reliability. In addition, personalized push notifications based on user profiles meet the specific needs of journalists, reduce search difficulty, and improve work efficiency. In summary, this solution brings a more efficient and accurate way to acquire media information to the news industry.

[0068] In one example, the process of scraping data from multiple data source platforms and constructing a dataset is described, such as... Figure 2 As shown, it includes:

[0069] Step 201: Use web crawling technology to extract source data from government websites, financial media, company websites, WeChat official accounts, and social media platforms.

[0070] In this example, web crawling technology collects data from web pages by simulating user behavior on the internet, such as browsing web pages and clicking links. The technology extracts the required source data information by parsing the HTML (Hypertext Markup Language) or JSON (JavaScript Object Notation) structures of web pages from multiple source data platforms, including government websites, financial media, company websites, public accounts, and social media platforms.

[0071] Step 202: Preprocess the source data to obtain the data, and use a big data processing framework and a distributed storage system to distribute and store the data to build a dataset.

[0072] In this example, the preprocessing of the source data includes: data cleaning, which involves removing duplicates, filling in missing values, and correcting erroneous data to ensure the accuracy and integrity of the data; data transformation, which involves adjusting the data format, standardizing the data, and performing necessary encoding conversions to unify the data format and facilitate subsequent processing; and data integration, which integrates information from different data sources, resolves data conflicts, and forms comprehensive and coherent data.

[0073] After preprocessing the source data, the resulting data is processed using big data processing frameworks such as Hadoop and Spark, and then stored in a distributed storage system, such as the Hadoop Distributed File System (HDFS). Distributed storage systems significantly increase storage capacity and enhance data fault tolerance and access speed by distributing data across multiple physical nodes.

[0074] In one example, a label classification model is used to classify the data in the dataset, obtaining at least one type label for each data point, including:

[0075] Step 301: Obtain the training dataset, which includes the news text data to be trained.

[0076] In this example, the process of obtaining the news text data to be trained from the training dataset is the same as the process of obtaining the data from the dataset. It is also collected from various news websites, social media, government announcements and other channels through web crawlers.

[0077] Step 302: Match at least one dimension of type label to the news text data to be trained based on the predefined multi-dimensional type label set.

[0078] In this example, at least one type label is matched for each piece of news text data to be trained, based on a predefined set of multi-dimensional type labels. Type labels can be based on the content of the news (e.g., "politics", "economics", "technology", etc.), sentiment (e.g., "positive", "negative", "neutral", etc.), or event type (e.g., "policy release", "market dynamics", "corporate announcements", etc.).

[0079] Step 303: Extract keywords and themes from the news text data to be trained to obtain content features.

[0080] In this example, an initial classification model is used to extract keywords and topics from the news text data to be trained. These keywords and topics are then used as content features input to the initial classification model for training.

[0081] Step 304: Train the initial classification model based on the content features and type labels of the news text data to be trained, and obtain the label classification model.

[0082] In this example, the content features and corresponding type labels of the news text data to be trained are used to train the initial classification model. Through iterative training and adjustment of model parameters, the label classification model can learn how to map the extracted content features of the news text data to the corresponding type labels, thereby improving classification accuracy.

[0083] Step 305: Use a label classification model to obtain at least one type label corresponding to the news text data in the dataset.

[0084] In this example, a trained label classification model is used to classify the news text data in the dataset to obtain at least one type label for each data point.

[0085] In one example, a scoring system is used to assess the quality of the data and obtain a quality score, including:

[0086] Step 401: Score the data from multiple dimensions and set corresponding weights for each dimension; the scoring dimensions include content score and influence score.

[0087] In this example, the data scoring dimensions include content scoring and impact scoring. Content scoring focuses on the quality of the data itself, such as accuracy, completeness, timeliness, relevance, and consistency. Impact scoring, on the other hand, assesses the data's influence and importance within its relevant domain or user group.

[0088] Each scoring dimension is assigned a weight based on the intended use of the data and the evaluation objectives. The weight reflects the importance of each dimension in the overall data quality assessment. For example, content scoring is more critical for some analyses and therefore receives a higher weight; while in other scenarios, influence scoring may be more important.

[0089] Step 402: Calculate the quality score of the data based on the rating and weight of each dimension of the data.

[0090] In this example, the score for each dimension is multiplied by its corresponding weight to obtain a weighted score for that dimension. The weighted scores for all dimensions are then summed to obtain the overall quality score. Based on this overall quality score, the data quality can be graded or categorized to guide recommendation decisions.

[0091] In one example, content scoring includes an assessment of the accuracy, depth, coherence, and originality of the data, including:

[0092] The accuracy assessment of the data includes: obtaining at least one reliable data source, integrating the text content that matches the data from the reliable data source, performing a logical consistency check on the data and the integrated text content, and obtaining a score for the accuracy of the data.

[0093] In this example, during the process of collecting data from data sources, such as data released by official sources and research reports from well-known research institutions, these data sources are more authoritative than other data sources, and therefore can be called trusted data sources.

[0094] The process involves identifying and integrating text content from trusted data sources that matches the data to be evaluated. Fact-checking algorithms (such as string matching and semantic similarity algorithms) are then used to logically infer information from the integrated text content to check for contradictions. Based on the calculated matching or similarity scores, an accuracy score is assigned to the data.

[0095] In-depth evaluation of data includes: calculating the information density and the range of topics covered by the data, and determining a score for the depth of the data.

[0096] In this example, information density refers to the amount of effective information contained per unit quantity (such as a unit of text) in the data, measuring the "value" or "information richness" of the data. Information density is obtained by calculating the frequency of keywords (i.e., the number of times important or related words appear in the data) or the complexity of sentences (i.e., the structure, length, and diversity of information contained in the sentences). High information density data means that a large amount of useful information is contained in a relatively small amount of text or data.

[0097] Thematic scope refers to the breadth and diversity of topics or concepts covered by data. A dataset may focus on a narrow but in-depth topic, or it may broadly cover multiple different topics. Thematic scope is determined by analyzing the number of all primary and secondary topics appearing in the data. A broad thematic scope means that the data covers multiple fields or perspectives, thus providing a more comprehensive view.

[0098] Based on the information density and the range of topics covered by the acquired data, the depth score of the data is determined.

[0099] The evaluation of data coherence involves training a deep learning network, including marginal loss functions and negative sampling, with both continuous and incoherent sentence pairs, and then using the deep learning network to obtain a score for the coherence of the data.

[0100] In this example, two sets of sentence pairs are prepared as training data: one set consists of logically coherent sentence pairs, and the other set consists of incoherent sentence pairs. These sentence pairs will be used to train a deep learning network that incorporates a marginal loss function and negative sampling techniques to learn how to distinguish between coherent and incoherent sentence pairs. By feeding the data to be evaluated into the trained deep learning network, a coherence score is obtained, which will reflect the degree of logical and semantic coherence of the data.

[0101] The evaluation of data originality includes: using a trained Word2Vec model to convert the text of each data point in the dataset into word vectors and aggregating them to obtain article vectors; calculating the similarity between two data points based on the article vectors; and determining the data originality score based on the similarity.

[0102] In this example, a pre-trained Word2Vec model is used to convert the text content of each data point in the dataset into word vectors. These word vectors capture the semantic relationships between words and map the text from a high-dimensional discrete space to a low-dimensional continuous vector space. Then, by aggregating the word vectors in the text (e.g., using average pooling), an article vector representing the features of the entire text is obtained. Next, the similarity between the article vectors of two data points (i.e., two articles) is calculated to measure their content similarity. The closer the similarity value is to 0, the greater the difference in content between the two articles, i.e., the higher the originality; conversely, a similarity value close to 1 indicates high content similarity, potentially indicating plagiarism or duplicate content. Finally, based on the similarity calculation results, an originality score is assigned to the data, which intuitively reflects the originality of the data.

[0103] In one example, the influence score includes: importance and real-time reach, including:

[0104] Assessing the importance of data involves obtaining the number of citations and media references to the data, and then determining a score to indicate the importance of the data based on these citation and media reference counts.

[0105] In this example, citation counts are a crucial indicator for measuring the academic impact and importance of data. If data is frequently cited in other academic articles, reports, or policy documents, it demonstrates that it contains valuable information or insights and has had a significant impact on academia or policymakers. Secondly, media citation counts reflect the data's influence and reach among the general public. If data is widely cited by mainstream media, well-known blogs, or websites, it is likely to have attracted widespread public attention and discussion, resulting in a significant social impact. After obtaining the citation counts and media citation counts of the data, these two indicators can be used to determine the data's importance score.

[0106] Predicting the real-time dissemination of data includes: obtaining interaction metrics of the data within a preset time period, and predicting the score of the real-time dissemination of the data based on the interaction metrics.

[0107] In this example, the interaction metrics for the data within a preset time period include likes, views, and shares. Based on these interaction metrics, a predictive model is built to evaluate the real-time dissemination of the data. When predicting the score, the model comprehensively considers factors such as likes, views, and shares, and weights them according to their contribution to dissemination. Finally, the model outputs a real-time dissemination score to measure the data's dissemination effect within the current time period.

[0108] In one example, using a government announcement titled "Adjustments to New Energy Vehicle Subsidy Policy" as an example, the process of quality assessment for the data quality scoring system is explained in detail:

[0109] First, assume that the rating dimensions and their corresponding weights are as shown in Table 1 below:

[0110]

[0111] Table 1. Scoring Dimensions and Corresponding Weight Allocation

[0112] The specific scoring process for the content of this announcement article is as follows:

[0113] Accuracy: Compared with the official website of the Ministry of Industry and Information Technology, the white paper of the Ministry of Finance, and industry association reports, the matching rate is 92%, and it is scored as 5 points. Calculation formula: 5 points × 30% = 1.5.

[0114] Depth: Information density = 0.85 (85 policy keywords per thousand words), covering three themes: subsidy standards, implementation time, and regulatory mechanisms, scored as 5 points. Calculation formula: 5 points × 25% = 1.25.

[0115] Coherence: After training with the BERT model, the announcement article was tested, and the context relevance score was 0.78, which was scored as 5 points. Calculation formula: 5 points × 25% = 1.25.

[0116] Originality: The text content of the announcement is converted into an article vector and compared with the database. The highest similarity score is 0.25, which is scored as 5 points. Calculation formula: 5 points × 20% = 1.0.

[0117] The specific process for scoring the impact of this announcement article is as follows:

[0118] Importance: Cited in 21 brokerage reports and 35 media reports, scored as 3 points. Calculation formula: 3 points × 50% = 1.5.

[0119] Dissemination: 12,000 views, 850 shares, 1,500 likes. Dissemination Index = 0.4 × 12,000 + 0.3 × 850 + 0.3 × 1,500 = 4,800 + 255 + 450 = 5,505, scored as 5 points. Calculation formula: 5 points × 50% = 2.5.

[0120] Total score calculation:

[0121] Content rating = 1.5 + 1.25 + 1.25 + 1.0 = 5.0 (out of 5);

[0122] Influence score = 1.5 + 2.5 = 4.0 (out of 5);

[0123] Total mass score = 5.0 × 60% + 4.0 × 40% = 3.0 + 1.6 = 4.6.

[0124] Based on the quality grading criteria shown in Table 2 below, this announcement article can be rated as Grade A:

[0125] Total score range Rating Processing strategy 4.5-5.0 Grade A Recommended to system users 3.5-4.4 Grade B Industry expert review is required. 2.0-3.4 Class C For auxiliary analysis only <2.0 Class D Automatic archiving does not participate in model training

[0126] Table 2 Quality Grading Standards

[0127] In one example, the process of obtaining user information and building a user profile, such as... Figure 5 As shown, it includes:

[0128] Step 501: Obtain the user's search history, behavioral characteristics, and points of interest to obtain the user information.

[0129] In this example, data related to user browsing and operation processes is crawled and collected, including: records left by users when searching on the Internet; user behavior patterns on websites or applications, such as clicks and browsing, and behavioral characteristics that reveal user preferences, activity levels, and possible purchase intentions; and identifying users' explicitly expressed or implied areas of interest, such as sports, music, and technology, through user browsing history, followed content, or social media activities.

[0130] Step 502: Based on the user information, determine the user's preference information and behavior patterns to obtain the user profile.

[0131] In this example, a deep neural network is used to extract features and perform pattern recognition on user information to identify user preferences and behavioral patterns. The user profile built on this basis is a comprehensive model that integrates static user attributes (such as age, gender, and region) and dynamic features (such as changes in interests and behavioral patterns).

[0132] This disclosure also provides a media information acquisition system, such as Figure 6 As shown, the system includes:

[0133] The crawling module 601 is used to crawl data from multiple data source platforms and build a dataset; the data is news text data.

[0134] Classification module 602 is used to classify the data in the dataset using a label classification model and obtain at least one type label for each data point.

[0135] The scoring module 603 is used to assess the quality of data using a scoring system and obtain a quality score for the data.

[0136] User profile building module 604 is used to obtain user information and build user profiles;

[0137] Feedback module 605 is used to match at least one target data that matches the user profile from the dataset based on the data's type label and quality score, and then provide feedback.

[0138] In one example, the crawling module 601 is specifically used for:

[0139] Web scraping technology was used to extract source data from government websites, financial media, company websites, WeChat official accounts, and social media platforms.

[0140] The source data is preprocessed to obtain the data, and a big data processing framework and a distributed storage system are used to distribute the data and build a dataset.

[0141] In one example, the classification module 602 is specifically used for:

[0142] Obtain the training dataset, which includes the news text data to be trained;

[0143] Based on a predefined set of multi-dimensional type labels, match at least one dimension of type label to the news text data to be trained;

[0144] Extract keywords and themes from the news text data to be trained to obtain content features;

[0145] Based on the content features and type labels of the news text data to be trained, the initial classification model is trained to obtain the label classification model;

[0146] Using a label classification model, obtain at least one type label corresponding to the news text data in the dataset.

[0147] In one example, the rating module 603 is specifically used for:

[0148] The data is scored across multiple dimensions, with corresponding weights assigned to each dimension; the scoring dimensions include content scoring and influence scoring.

[0149] Calculate the data quality score based on the rating and weight of each dimension of the data.

[0150] In one example, content scoring includes evaluating the accuracy, depth, coherence, and originality of the data. Scoring module 603 is specifically used in content scoring to:

[0151] The accuracy assessment of the data includes: obtaining at least one reliable data source, integrating the text content that matches the data from the reliable data source, performing a logical consistency check on the data and the integrated text content, and obtaining a score for the accuracy of the data.

[0152] In-depth evaluation of data includes: calculating the information density and the range of topics covered by the data, and determining the depth score of the data;

[0153] The evaluation of data coherence includes: training a deep learning network with marginal loss function and negative sampling using continuous sentence pairs and incoherent sentence pairs, and using the deep learning network to obtain a score of data coherence;

[0154] The evaluation of data originality includes: using a trained Word2Vec model to convert the text of each data point in the dataset into word vectors and aggregating them to obtain article vectors; calculating the similarity between two data points based on the article vectors; and determining the data originality score based on the similarity.

[0155] In one example, the influence score includes: importance and real-time reach. The scoring module 603 is specifically used for:

[0156] Assessing the importance of data includes obtaining the number of citations and media references to the data, and determining the importance score of the data based on the number of citations and media references.

[0157] Predicting the real-time dissemination of data includes: obtaining interaction metrics of the data within a preset time period, and predicting the score of the real-time dissemination of the data based on the interaction metrics.

[0158] According to embodiments of this disclosure, this disclosure also provides an electronic device and a readable storage medium.

[0159] Figure 7A schematic block diagram of an example electronic device 800 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0160] like Figure 7 As shown, the electronic device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. The RAM 803 may also store various programs and data required for the operation of the device 800. The computing unit 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0161] Multiple components in device 800 are connected to I / O interface 805, including: input unit 806, such as keyboard, mouse, etc.; output unit 807, such as various types of monitors, speakers, etc.; storage unit 808, such as disk, optical disk, etc.; and communication unit 809, such as network card, modem, wireless transceiver, etc. Communication unit 809 allows device 800 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0162] The computing unit 801 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 801 performs the various methods and processes described above, such as the media information acquisition method. For example, in some embodiments, the media information acquisition method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on device 800 via ROM 802 and / or communication unit 809. When the computer program is loaded into RAM 803 and executed by the computing unit 801, one or more steps of the media information acquisition method described above may be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to perform the media information acquisition method by any other suitable means (e.g., by means of firmware).

[0163] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0164] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0165] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0166] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0167] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0168] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0169] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0170] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this disclosure, "a plurality of" means two or more, unless otherwise explicitly specified.

[0171] The above description is merely a specific embodiment of this disclosure, but the scope of protection of this disclosure is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this disclosure should be included within the scope of protection of this disclosure. Therefore, the scope of protection of this disclosure should be determined by the scope of the claims.

Claims

1. A media information acquisition method characterized by comprising: The method comprises: grabbing data from multiple data source platforms to build a data set; the data is news text data; classifying the data in the data set using a label classification model to obtain at least one type label corresponding to each data; using a scoring system to evaluate the quality of the data and obtain a quality score of the data, including: scoring the data in multiple dimensions and setting corresponding weights for each dimension; the scoring dimensions include content score and influence score; according to the score and weight of each dimension of the data, the quality score of the data is calculated; the content score includes evaluating the accuracy, depth, coherence and originality of the data, including: the accuracy evaluation of the data includes: obtaining at least one trusted data source, integrating the text content in the trusted data source that matches the data, and checking the logical consistency of the data and the integrated text content to obtain the accuracy score of the data; the depth evaluation of the data includes: calculating the information density and the range of covered topics of the data to determine the depth score of the data; the coherence evaluation of the data includes: training a deep learning network including an edge loss function and negative sampling using continuous sentence pairs and incoherent sentence pairs, and using the deep learning network to obtain the coherence score of the data; the originality evaluation of the data includes: converting the text of each data in the data set into a word vector using a trained Word2Vec model and aggregating to obtain an article vector of the data, calculating the similarity between two data according to the article vector, and determining the originality score of the data according to the similarity; the influence score includes importance and real-time propagation, including: the importance evaluation of the data includes: obtaining the number of citations and the number of media references of the data, and determining the importance score of the data according to the number of citations and the number of media references; the real-time propagation prediction of the data includes: obtaining the interaction indicators of the data within a preset time period, and predicting the real-time propagation score of the data according to the interaction indicators; obtaining user information to build a user portrait; According to the type label and quality score of the data, at least one target data meeting the user portrait is matched from the data set and feedback is given.

2. The method of claim 1, wherein, The method comprises: using crawler technology to grab source data from government websites, financial media, company websites, public accounts and social media platforms; preprocessing the source data to obtain the data, and using a big data processing framework and a distributed storage system to store the data in a distributed manner to build the data set.

3. The method of claim 1, wherein, The method comprises: obtaining a training data set, which includes news text data to be trained; According to a pre-defined multi-dimensional type label set, at least one dimensional type label is matched for the news text data to be trained; extract keywords and topics in the to-be-trained news text data to obtain content features; train an initial classification model according to the content features and type labels of the to-be-trained news text data to obtain a label classification model; acquire at least one type label corresponding to the news text data in the data set by using the label classification model.

4. The method of claim 1, wherein, The user information is acquired, and the user portrait is constructed, including: acquiring search history, behavior characteristics and interest points of a user to obtain the user information; determining preference information and behavior patterns of the user according to the user information to obtain the user portrait.

5. A media information acquisition system characterized by comprising: The system comprises: a crawling module configured to crawl data from a plurality of data source platforms to construct a data set; the data is news text data; a classification module configured to classify data in the data set by using a label classification model to acquire at least one type label corresponding to each data; a scoring module configured to perform quality evaluation on the data by using a scoring system to acquire a quality score of the data, including: performing multi-dimensional scoring on the data, and setting a corresponding weight for each dimension; the scoring dimensions include content scoring and influence scoring; and calculating the quality score of the data according to the score and weight of each dimension of the data; the content scoring includes evaluating the accuracy, depth, coherence and originality of the data, and the scoring module is specifically configured to: the accuracy evaluation on the data includes: acquiring at least one trusted data source, integrating text content in the trusted data source that matches the data, and performing logical consistency checking on the data and the integrated text content to obtain the accuracy score of the data; the depth evaluation on the data includes: calculating the information density and the covered topic range of the data to determine the depth score of the data; the coherence evaluation on the data includes: training a deep learning network including an edge loss function and negative sampling by using continuous sentence pairs and incoherent sentence pairs, and acquiring the coherence score of the data by using the deep learning network; the originality evaluation on the data includes: converting the text of each data in the data set into a word vector and aggregating by using a trained Word2Vec model to obtain an article vector of the data, calculating the similarity between two data according to the article vector, and determining the originality score of the data according to the similarity; the influence scoring includes importance and real-time propagation, and the scoring module is specifically configured to: the importance evaluation on the data includes: acquiring the number of citations and the number of media references of the data, and determining the importance score of the data according to the number of citations and the number of media references; the real-time propagation prediction of the data includes: acquiring interaction indicators of the data within a preset time period, and predicting the real-time propagation score of the data according to the interaction indicators; a user portrait construction module configured to acquire user information and construct a user portrait; a feedback module configured to match at least one target data from the data set that matches the user profile according to the type label and quality score of the data and to provide feedback.

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

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

Images