WO2020155496A1 - Public opinion tracking method and device for combined video-text data, and computer apparatus - Google Patents

Abstract

The present application provides a public opinion tracking method and device for combined video-text data, and a computer apparatus. The method comprises: obtaining specified public data, and building a public database; filtering the public database to obtain public opinion combined video-text data, public opinion text data, and public opinion video data, all corresponding to a public opinion topic; and combining all of the data obtained by filtering so as to form public opinion data. The present application performs tracking by combining text and video in public opinion information, thereby achieving comprehensive tracking of public opinion information.

Description

Public opinion tracking method, device and computer equipment for video text combined data

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on January 31, 2019, the application number is 201910100413.X, and the invention title is "Public Opinion Tracking Method, Apparatus and Computer Equipment for Video Text Combined Data", all of which The content is incorporated in this application by reference.

Technical field

This application relates to the field of data analysis technology, and in particular to a public opinion tracking method, device and computer equipment for video text combined data.

Background technique

The current methods of monitoring public opinion on the Internet use keyword matching or text OCR recognition technology to capture relevant news, social dynamics, and netizen comments on various platforms on the Internet. In this case, only the data containing the relevant keyword text will be processed, and the volume is only the volume of the text, and the content that only contains the video without the relevant keyword cannot be searched. Now more and more users like to use videos to express their emotions. They may not mention certain keywords, but they use videos to convey the same information. For example, there are millions of short videos posted daily by real users of Weibo. At present, due to the lack of effective identification methods, video data is a long-standing blind spot for monitoring.

technical problem

The main purpose of this application is to provide a public opinion tracking method, device and computer equipment for video text combined data, which aims to solve the disadvantages of the existing public opinion monitoring methods that cannot achieve public opinion tracking on video.

Technical solutions

To achieve the above objective, this application provides a public opinion tracking method for video text combined data, including:

According to a first preset frequency, obtain designated public data from a first preset network platform, the designated public data including all the public information of the first preset network platform and the propagation path corresponding to each public information , The public information includes individual text information with only text, individual video information with only video, and combined video and text information related to the video and the text;

Construct a public database according to the specified public data;

Screening and obtaining public opinion video text combination data corresponding to the preset public opinion topic in the public database;

According to the text similarity, the public opinion text data whose similarity with the text data of the public opinion video text combined data is higher than the first preset value is filtered in the public database; and according to the video source address and the video similarity, The public opinion video data whose similarity with the video data of the public opinion video text combined data is higher than a second preset value is filtered from the public database;

Obtaining public opinion data according to the public opinion video text combination data, the public opinion text data, and the public opinion video data;

According to the public opinion data, the popularity trend of the preset public opinion topic is obtained.

This application also provides a public opinion tracking device for video text combined data, including:

The first obtaining module is configured to obtain designated public data from the first preset network platform according to the first preset frequency;

A building module for building a public database based on the designated public data;

The first screening module is configured to filter the public opinion video text combination data corresponding to the preset public opinion topic in the public database;

The second filtering module is configured to filter the public opinion text data with the text data of the public opinion video text combined data that is higher than the first preset value in the public database according to the text similarity; and according to the video source Address and video similarity, filtering the public opinion video data with the video data of the public opinion video text combination data that has a similarity higher than a second preset value from the public database;

The first generating module is configured to obtain public opinion data according to the public opinion video text combination data, the public opinion text data, and the public opinion video data;

The analysis module is used to obtain the trend of the hotness change of the preset public opinion topic according to the public opinion data.

The present application also provides a computer device, including a memory and a processor, the memory stores computer-readable instructions, and the processor implements the steps of any one of the foregoing methods when the computer-readable instructions are executed by the processor.

The present application also provides a computer non-volatile readable storage medium, on which computer readable instructions are stored, and when the computer readable instructions are executed by a processor, the steps of any one of the above methods are implemented.

Beneficial effect

The public opinion tracking method, device and computer equipment of video text combined data provided in this application realizes comprehensive coverage of public opinion information by combining text and video in public opinion information for tracking, and effectively increases the popularity of public opinion information Resolution accuracy of changing trends.

Description of the drawings

Figure 1 is a schematic diagram of the steps of a public opinion tracking method for video text combined data in an embodiment of the present application;

2 is a block diagram of the overall structure of a public opinion tracking device for video text combined data in an embodiment of the present application;

FIG. 3 is a schematic block diagram of the structure of a computer device according to an embodiment of the present application.

The realization, functional characteristics, and advantages of the purpose of this application will be further described in conjunction with the embodiments and with reference to the drawings.

The best mode of the invention

In order to make the objectives, technical solutions, and advantages of this application clearer, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the application, and not to limit the application.

1, an embodiment of the present application provides a public opinion tracking method for video text combined data, including:

S1: According to the first preset frequency, obtain designated public data from the first preset network platform. The designated public data includes all public information of the first preset network platform and the propagation path corresponding to each public information. The public information includes Individual text information with only text, individual video information with only video, and video text combined information associated with video and text;

S2: construct a public database according to the specified public data;

S3: Filter the public opinion video text combination data corresponding to the preset public opinion topic in the public database;

S4: According to the text similarity, filter in the public database to obtain the public opinion text data whose similarity with the text data of the public opinion video text combination data is higher than the first preset value; and according to the video source address and the video similarity , Screening in the public database to obtain public opinion video data whose similarity with the video data of the public opinion video text combined data is higher than a second preset value;

S5: Obtain public opinion data based on the combined public opinion video text data, public opinion text data, and public opinion video data;

S6: According to the public opinion data, obtain the trend of change in the popularity of the preset public opinion topic.

In this embodiment, the public opinion system is pre-associated with the first preset network platform according to the developer's setting, so as to obtain information publicly released by the users of each platform in the first preset network platform. Among them, the first preset network platform is a communication path through which publicly released information can be queried, and the publicly released information includes a video network platform, such as online social platforms such as Weibo and Douyin. The first preset network platform can directly query the information publicly released by the users of each platform, as well as the relevant interactive actions such as comments, forwarding, or likes corresponding to the information, and query all the communication paths corresponding to the information according to these interactive actions. The public opinion system can be associated with multiple first preset network platforms at the same time to obtain publicly released information and integrate and analyze it. When the public opinion system is in use, according to the developer's setting, according to the first preset frequency, the public data used in the time period is directly obtained from the first preset network platform through the web crawler, that is, the designated public data. Among them, the designated public data includes the public information publicly released by all platform users during the time period and the respective propagation paths of these public information. The public information in the designated public data includes separate text information with only text, separate video information with only video, and combined video text information published in association with video and text. For example, a platform user attaches a video when publishing a text, which is Video text combination information. The public opinion system builds a public database to store all designated public data. The public opinion system has preset public opinion topics. Among them, the preset public opinion topic can be a single word or a sentence. The preset public opinion topic can be manually input by the developer, or it can be obtained by the public opinion system itself according to a preset setting. For example, the public opinion system monitors the hot topic lists of social platforms such as Weibo and Douyin, and selects the top topic of the topic list as the preset public opinion topic. The public opinion system inputs the preset public opinion topics into the public database, first selects the first public opinion keywords from the preset public opinion topics according to the part of speech, and filters out the text information from the video text combination sub-database according to the first public opinion keywords. Part of the video text combined data of a public opinion keyword is used as the video text combined sub-data. Then, the same videos in the video text combination sub-data, that is, the number of appearances of each video sub-data are counted. If the number of occurrences of a certain video sub-data is greater than the preset number, the public opinion system determines that the video sub-data is public opinion video sub-data. The public opinion system obtains public opinion video text combination data related to the public opinion topic according to the public opinion video sub-data and the text data corresponding to the public opinion video sub-data. Based on the combined data of the public opinion video text, the public opinion system obtains the original source address of each individual video data and public opinion video sub-data, that is, the video source address, and then separates the original source address of each individual video data with the original public opinion video sub-data. The source addresses are compared, and the individual video data with the same original source address is used as the first video data, and the individual video data with the original source address inconsistent is used as the second video data. Then, the public opinion system uses the public opinion video sub-data in the public opinion video text combination data as a benchmark, and according to the video similarity, filters the second video data from the second video data that has a video similarity value greater than the second preset value. The video data serves as the third video data. The public opinion system integrates the first video data and the second video data to obtain public opinion video data. At the same time, the public opinion system compares the public opinion text sub-data in the combined public opinion video text data with the individual text data, and selects keywords and part-of-speech analysis to filter the public opinion text sub-data from the separate text database. A preset value, for example, part of individual text data with a similarity of more than 90% is used as individual public opinion text data. The public opinion system aggregates individual public opinion text data, individual public opinion video data, and public opinion video text combined data to generate public opinion data. The public opinion system can obtain the trend of the popularity of the preset public opinion topic through the analysis of the public opinion data, specifically: the public opinion data includes the number of users of all public opinion content about the preset public opinion topic released within the time period, and comments related to the public opinion content According to the number of users, the number of comments, and the number of reposts, the public opinion system can determine the popularity of the topic of public opinion, and at the same time, it can know that the topic of public opinion is in the region based on the user’s location information. On the degree of diffusion. The public opinion system comprehensively pays attention to the degree of popularity and geographical spread, and can intuitively analyze the trend of the preset public opinion topic. The analysis process is specifically as follows: the public opinion data includes the number of users of all public opinion content about the preset public opinion topics released within the time period, the number of comments related to the public opinion content, the forwarding volume related to the public opinion content, and related user location information. The numerical value of the number of users, the number of comments, and the amount of forwarding can determine the popularity of the public opinion topic, and at the same time, the extent of the geographical spread of the public opinion topic can be obtained according to the user location information.

Further, the public database includes a separate text sub-database, a separate video sub-database, and a video text combined sub-database, the separate text sub-database is a database composed of multiple sets of separate text data, and the separate video sub-database is multiple sets A database composed of separate video data, the video text combined sub-database is a database composed of multiple sets of video and text one-to-one correspondingly associated data; the public opinion corresponding to the preset public opinion topic is obtained by screening in the public database The steps of combining video text data include:

S301: In the video text combination sub-database, filter the public opinion video text combination data according to the keyword of the preset public opinion topic, wherein each of the public opinion video text combination data consists of a piece of public opinion video subdata and One piece of public opinion text sub-data corresponds to the association composition;

According to the text similarity, the public opinion text data whose similarity with the text data of the public opinion video text combined data is higher than a first preset value is filtered in the public database; and according to the video source address and the video similarity The step of screening the public opinion video data with the video data of the public opinion video text combination data whose similarity is higher than a second preset value in the public database includes:

S401: In the separate text sub-database, filter according to the text similarity to obtain the public opinion text data whose similarity to the public opinion text sub-data is higher than the first preset value, and store it in the separate video sub-database , Filtering according to the video source address and video similarity to obtain public opinion video data whose similarity with the public opinion video sub-data is higher than the second preset value.

In this embodiment, the video information and text information in the video text combination information acquired by the public opinion system correspond to each other when the platform user discloses, that is, when the user discloses the text information, the video information published together in the same piece of information is attached. The public database constructed by the public opinion system includes a separate text sub-database, a separate video sub-database and a video text combined sub-database. Among them, the separate text sub-database is a database composed of multiple sets of separate text data, the separate video sub-database is a database composed of multiple sets of separate video data, and the video text combined sub-database is composed of multiple sets of videos and texts in one-to-one correspondence. Database. Each group of data includes public information and the corresponding transmission channels of public information. The public opinion system inputs the preset public opinion topic into the video text combination sub-database, and first parses the preset public opinion topic to obtain at least one first public opinion keyword, that is, the keyword of the preset public opinion topic. Then, each first public opinion keyword is input into the video text combination sub-database, and part of the video text combination data containing the first public opinion keyword in the text information is filtered out, and then the same video in the partial video text combination data is classified Class, and count the number of appearances of the same video. If the number of appearances of the same video is greater than the preset number of viewing times, the public opinion system determines that the same video is public opinion video sub-data. The public opinion system integrates public opinion video sub-data and public opinion text data corresponding to the public opinion video sub-data to generate public opinion video text combined data. Among them, the public opinion video text data includes public opinion video sub-data and public opinion text sub-data, and there is a one-to-one correspondence between the two, and an index relationship is constructed so that the two are related to each other according to the index relationship. For example, text A corresponds to video a, and the index of both is 1; text B corresponds to video b, and the index between the two is 2. After the public opinion system obtains the public opinion video text combination data, the public opinion system uses the public opinion video sub-data in the public opinion video text combined data as a benchmark, compares each individual video data with the public opinion video sub-data one by one, and filters to obtain the same original source At least one individual video data whose address or video similarity value is greater than the second preset value is used as public opinion video data. Since the individual public opinion video data is similar to the public opinion video data, the individual public opinion video data corresponds to the preset public opinion topic. At the same time, the public opinion system is based on the public opinion text sub-data in the public opinion video text combined data, through word embedding and part-of-speech analysis, from the public opinion text sub-data to filter out the second public opinion keywords corresponding to each text, and at the same time to filter out the separate text data Separate keywords for each text. Then, the public opinion system counts the number of times for each current public opinion keyword and individual keywords. If there are more occurrences of the same keyword in two texts, it means that the similarity of the two texts is higher. The public opinion system judges the similarity between the two texts based on this, and selects the part of the individual text data whose similarity with the text information in the combined public opinion video text data is greater than the first preset value from the individual text data as the public opinion text data.

Further, in the video text combination sub-database, the step of obtaining the public opinion video text combination data by filtering according to the keywords of the preset public opinion topic includes:

S3011: Parse the preset public opinion topic to obtain at least one first public opinion keyword;

S3012: In the video text combination sub-database, filter to obtain at least one set of video text combination data containing the first public opinion keyword as video text combination sub-data, and the video text combination sub-data consists of text sub-data and video sub-data. Data corresponding composition;

S3013: Calculate the number of appearances of each of the video sub-data separately;

S3014: Retrieve a preset number of times, compare each of the number of appearances with the preset number of times, and select the video sub-data corresponding to the number of appearances greater than the preset number of times as the public opinion video sub-data;

S3015: From the video text combination sub-database, respectively filter the text sub-data corresponding to each of the public opinion video sub-data as the public opinion text sub-data;

S3016: Correspond each of the public opinion text sub-data and each of the public opinion video sub-data respectively to obtain the public opinion video text combined data.

In this embodiment, based on the part of speech, the public opinion system performs analysis such as word segmentation and removal of stop words on a preset public opinion topic to obtain at least one first public opinion keyword. The video text combination sub-database includes multiple sets of data in one-to-one correspondence between video data and text data. In the public opinion system, the first public opinion keyword belongs to the video text combined sub-database, and at least one set of text data containing the first public opinion keyword is selected as the text sub-data from multiple sets of text sub-data. Further, in this embodiment, the number of occurrences of the first public opinion keyword of each text in the text sub-data can be counted, and part of the text sub-data whose occurrence number exceeds a preset number can be selected to improve the text sub-data and public opinion. The relevance of the topic. Since the text data corresponds to the video data, the public opinion system can directly filter out all the video sub-data corresponding to the text sub-data from the video data. The public opinion system separately counts the number of appearances of each video sub-data, then retrieves the preset number of times set in advance, and compares the number of appearances corresponding to each video sub-data with the preset number one by one, and selects the number of appearances greater than the preset number The corresponding video sub-data serves as the public opinion video sub-data. Finally, the public opinion system again filters out the public opinion text sub-data corresponding to the public opinion video sub-data from each group of text sub-data according to the correspondence between the video data and the text data. Public opinion summarizes all public opinion video sub-data and public opinion text sub-data, and associates them one by one according to the corresponding relationship to generate public opinion video text combination data.

Further, the step of respectively matching each of the public opinion text sub-data and each of the public opinion video sub-data to obtain the public opinion video text combined data includes:

S30161: Obtain the public address of each public opinion text sub-data and the public address of each public opinion video sub-data respectively;

S30162: Associate and group the public opinion text sub-data and public opinion video sub-data with the same public address, so as to achieve a one-to-one correspondence between each public opinion text sub-data and each public opinion video sub-data;

S30163: Obtain the public opinion video text combined data according to each of the public opinion text sub-data and each of the public opinion video sub-data after being associated and grouped.

In this embodiment, the public opinion system separately obtains the public address of each public opinion text sub-data and the public address of each public opinion video sub-data, where the public address is the original network address that discloses the data. The public opinion system distinguishes the correspondence between each public opinion text sub-data and each public opinion video sub-data according to the public opinion of the data, that is, two data with the same public address are information released together by the same user when they are public. The public opinion system associates public opinion text sub-data and public opinion video sub-data with the same public address into one group. Further, if there are multiple public opinion text sub-data and multiple public opinion video sub-data with the same public address, the public opinion system obtains the publication time of each data, and combines the public opinion text sub-data and public opinion video according to the same public address and the same public time. The sub-data are related to each other. After the public opinion system associates each public opinion text sub-data and each public opinion video sub-data into groups, the combined public opinion video text data is obtained.

Further, the separate video sub-database includes multiple sets of separate video data, and in the separate video sub-database, the similarity with the public opinion video sub-data is higher than that of the public opinion video sub-data by filtering according to the video source address and video similarity. The step of describing the public opinion video data of the second preset value includes:

S4011: Obtain the original source addresses of each of the individual video data and the preset video sub-data respectively;

S4012: Compare the original source addresses of the individual video data with the original source addresses of the public opinion video sub-data one by one, select the individual video data with the same original source address as the first video data, and select the original source address Inconsistent individual video data is used as the second video data;

S4013: Calculate video similarity values between each of the second video data and the public opinion video sub-data;

S4014: Compare each of the video similarity values one by one with the second preset value, and select the second video data corresponding to the video similarity value greater than the second preset value as the third video data;

S4015: Use the first video data and the third video data as public opinion video data.

In this embodiment, the designated public data also includes the original source address of each public information. The individual video sub-database is composed of multiple sets of individual video data. The public opinion system first obtains the original source address of each public opinion video sub-data from the public database, and constructs the original source address database according to the original source address of each public opinion video sub-data. Then obtain the original source address of the individual video data, and compare the original source address of the individual video data with each original source address in the original source address database based on each original source address in the original source address database. Determine whether the original source address of the individual video data is included in the original source address database. If it is included in the original source address database, it means that the individual video has the same original source address as the public opinion video and belongs to the same video, that is, the individual video data with the same original source address is selected as the first video data. If it does not have the same original source address as the public opinion video sub-data, it is regarded as the second video data. The public opinion system selects a preset number of play frame pictures from each public opinion video sub-data according to the play time of the video, and groups the preset number of play frame pictures and corresponding public opinion video sub-data into corresponding groups. The public opinion video sub-data uses a clustering model for each group of broadcast frame pictures, such as the DBSCAN clustering model, to form each group, and then randomly select a broadcast frame picture from each group as the key to the corresponding public opinion video sub-data Frame picture. The public opinion system groups all the key frame pictures according to the corresponding public opinion video sub-data to build a key frame picture library. For example, the key frame picture library contains: public opinion video A, key frame pictures: a1, b2, c3; public opinion video B, key frame pictures: a2, b3, c4. After constructing the key frame picture library, the public opinion system selects the key frame pictures of the individual video from the individual video data according to the above steps, and compares the key frame pictures of the individual video with the key frame picture library one by one. According to the same or similar number of key frame pictures of a single video and key frame pictures of a single group of public opinion video sub-data, divided by the total number of key frame pictures, the video similarity value between the two videos can be calculated. For example, if the key frame pictures of the sub-data of the individual video and the public opinion video are both 5, where there are 3 identical or similar key frame pictures, the video similarity value of the two videos is 60%. The same or similar judgment between the key frame pictures can be clustered through the DBSCAN clustering model. If two key frame pictures exist in the same group after clustering, the two pictures are the same or similar pictures. The public opinion system compares each video similarity value with the preset second preset value one by one, and uses the second video corresponding to the video similarity value greater than the second preset value as the third video. That is, if the video similarity value between the two videos is greater than the second preset value, the public opinion system determines that the two videos are the same or similar, and selects the second video data as the third video data. Wherein, the second preset value is preferably 80%. The public opinion system aggregates the first video data and the third video data and sets them as public opinion video data.

Further, in the separate text sub-database, the step of filtering according to text similarity to obtain public opinion text data whose similarity with the public opinion text sub-data is higher than the first preset value includes:

S4016: According to the part of speech, analyze each of the public opinion text sub-data and each of the individual text data to obtain a preset number of second public opinion keywords corresponding to each of the public opinion text sub-data, and the corresponding individual text data The preset number of individual keywords;

S4017: Compare each of the second public opinion keywords with each of the individual keywords, and filter to obtain the same keywords;

S4018: Count the number of appearances corresponding to each of the same keywords respectively;

S4019: Compare each of the occurrence times with a first threshold value one by one, and select the same keywords corresponding to the occurrence times greater than the first threshold value as designated keywords;

S40110: Select the individual text data containing the specified keywords as the public opinion text data.

In this embodiment, the public opinion text sub-data includes multiple public opinion texts, and the separate text database includes multiple individual texts. The public opinion system first performs word segmentation and removal of stop words on each public opinion text and individual text based on part of speech, and obtains a public opinion vocabulary corresponding to a single public opinion text and a separate vocabulary corresponding to a single individual text. Among them, word segmentation refers to the operation of decomposing a text into individual words such as subject, predicate, and object. After word segmentation, a corresponding relationship is established based on the original expression relationship of the subject, predicate, and object in the text. For example, in the sentence "I went to Beijing", the subject is "I", the predicate is "Go", and the object is "Beijing". After word segmentation, the three words are related according to the original order in the text. When the subject "I" is set as a keyword, the predicate "Go" or the object "Beijing" in the same sentence will also be related to it Together, make a combination. And to remove stop words is to remove meaningless words, such as "ah" and "oh". The public opinion system separately counts the word frequency of each word in the public opinion lexicon in the corresponding text. Then, the public opinion system separately calculates the reverse file frequency of each word, that is, the number of all texts corresponding to all words in the public opinion lexicon divided by the number of texts containing the word, and then the obtained quotient is obtained by taking the logarithm. The weight of each word in a single text is obtained by multiplying the word frequency of the word by the reverse document frequency. The importance of a single word increases in proportion to the number of times it appears in the document, but at the same time it decreases in inverse proportion to the frequency of its appearance in the corpus, that is, the greater the weight, the higher the importance. The public opinion system selects a preset number of words in the calculated single text in descending order of weight as the second public opinion keyword. At the same time, the public opinion system calculates the weight of each word in the separate thesaurus according to the above method, and filters from the separate thesaurus to the same preset number of separate keywords in descending order. The public sentiment system is based on the second public opinion keyword, selects the same keywords with the same second public opinion keyword from the individual keywords, and counts the number of times each same keyword appears in the corresponding single text In the text data, the same keywords whose appearance times are greater than the first threshold are filtered, and the individual text data corresponding to the same keywords are used as public opinion text data.

Further, before the step of obtaining designated public data from the first preset network platform according to the first preset frequency, the method includes:

S7: Acquire multiple sets of search data from a second preset network platform according to a second preset frequency, where the search data includes search information and the number of searches corresponding to the search information;

S8: Compare each number of searches with a second threshold respectively, and use search information corresponding to the number of searches greater than the second threshold as the preset public opinion topic.

In this embodiment, the public opinion topic may be manually input by the developer, or may be automatically screened by the public opinion system. The public opinion system is associated with a second preset network platform in advance, and the second preset network platform is a search platform, such as Baidu and Sogou search. According to the second preset frequency, the public opinion system obtains all search data in the time period from the last acquisition time to the current time from the second preset network platform. The search data includes the search information input by the user and the total number of searches corresponding to the search information. The public opinion system calls the second preset times, and compares the search times corresponding to each search information with the second preset times respectively, and filters out the designated search data whose search times are greater than the second preset times from the search data. The public opinion system automatically sets the search information in the specified search data as public opinion topics.

Further, the designated public data includes the public time of the information, the public opinion system is pre-associated with at least one preset terminal, and after the step of obtaining the trend of the preset public opinion topic based on the public opinion data, include:

S9: Determine whether the heat change trend meets the conditions for triggering automatic push;

S10: If the conditions for triggering automatic push are met, filter the public opinion data to obtain the designated push public opinion data, and the push public opinion data is the data with the shortest time of publication from the current time;

S11: Obtain the link of the pushed public opinion data, and generate push information including the link of the pushed public opinion data;

S12: Send the push information to each of the preset terminals respectively.

In this embodiment, the public opinion system analyzes the public opinion data, and can obtain the trend of the popularity of the public opinion topic. The public opinion system comprehensively pays attention to the degree of popularity and geographical spread, and can intuitively obtain the trend of the preset public opinion topic. The public opinion system judges whether the trend of public opinion topics meets the conditions for triggering automatic push according to the degree of attention and geographical spread in the trend of heat changes. If the focus of attention in the popularity trend exceeds the preset number, and the spread of the region exceeds the preset range, such as the number of reposts and comments are greater than 500,000, and the spread of the region exceeds 100,000 square kilometers, then judge the topic of public opinion The heat change trend meets the conditions for triggering automatic push. The public opinion system screens out the public opinion video text combination data that includes text and video from the public opinion data, and at the same time, the public opinion video text data with the shortest time interval between the public release and the current time is used as the push public opinion data. The public opinion system obtains the link to push the public opinion data, generates push information containing the link to push the public opinion data, and then automatically sends the push information to the preset terminal so that the public can understand the current public opinion topic in time.

The public opinion tracking method of video text combined data provided in this embodiment realizes comprehensive coverage of public opinion information by combining text and video in public opinion information for tracking, and effectively improves the accuracy of the analysis of the trend of public opinion information. degree.

Referring to Figure 2, an embodiment of the present application also provides a public opinion tracking device for video text combined data, including:

The first obtaining module 1 is configured to obtain designated public data from a first preset network platform according to a first preset frequency;

The construction module 2 is used to construct a public database according to the specified public data;

The first screening module 3 is configured to filter the public opinion video text combination data corresponding to the preset public opinion topic in the public database;

The second screening module 4 is configured to filter the public opinion text data whose similarity with the text data of the public opinion video text combination data is higher than the first preset value in the public database according to the text similarity; and according to the video The source address and the video similarity, and the public opinion video data whose similarity with the video data of the public opinion video text combined data is higher than a second preset value is obtained by screening in the public database;

The first generating module 5 is configured to obtain public opinion data according to the public opinion video text combination data, the public opinion text data, and the public opinion video data;

The analysis module 6 is configured to obtain the trend of the popularity change of the preset public opinion topic according to the public opinion data.

In this embodiment, the embodiments of the modules of the public opinion tracking device are consistent with the corresponding method steps described above, and will not be described in detail here.

The public opinion tracking device for video text combined data provided in this embodiment realizes comprehensive coverage of public opinion information by combining text and video in public opinion information for tracking, and effectively improves the accuracy of analyzing the trend of public opinion information. degree.

Referring to FIG. 3, an embodiment of the present application also provides a computer device. The computer device may be a server, and its internal structure may be as shown in FIG. 3. The computer equipment includes a processor, a memory, a network interface and a database connected by a system bus. Among them, the computer designed processor is used to provide calculation 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, computer readable instructions, and a database. The internal memory provides an environment for the operation of the operating system and computer-readable instructions in the non-volatile storage medium. The database of the computer equipment is used to store data such as public databases. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer-readable instructions are executed by the processor, the processes of the foregoing method embodiments are executed. Those skilled in the art can understand that the structure shown in FIG. 3 is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied.

An embodiment of the present application further provides a computer non-volatile readable storage medium, on which computer readable instructions are stored, and when the computer readable instructions are executed by a processor, the processes of the above-mentioned method embodiments are executed. The above are only the preferred embodiments of this application, and do not limit the scope of this application. Any equivalent structure or equivalent process transformation made using the content of the description and drawings of this application, or directly or indirectly applied to other related The technical field is equally included in the scope of patent protection of this application.

Claims (20)

1. A public opinion tracking method for video text combined data, which is characterized in that it comprises:

   According to a first preset frequency, obtain designated public data from a first preset network platform, the designated public data including all the public information of the first preset network platform and the propagation path corresponding to each public information , The public information includes individual text information with only text, individual video information with only video, and combined video and text information related to the video and the text;

   Construct a public database according to the specified public data;

   Screening and obtaining public opinion video text combination data corresponding to the preset public opinion topic in the public database;

   According to the text similarity, the public opinion text data whose similarity with the text data of the public opinion video text combined data is higher than the first preset value is filtered in the public database; and according to the video source address and the video similarity, The public opinion video data whose similarity with the video data of the public opinion video text combined data is higher than a second preset value is filtered from the public database;

   Obtaining public opinion data according to the public opinion video text combination data, the public opinion text data, and the public opinion video data;

   According to the public opinion data, the popularity trend of the preset public opinion topic is obtained.
2. The method for tracking public opinion of video text combined data according to claim 1, wherein the public database includes a single text sub-database, a single video sub-database, and a video text combined sub-database, and the single text sub-database is a group of multiple A database composed of separate text data, the separate video sub-database is a database composed of multiple sets of separate video data, and the video-text combination sub-database is a database composed of multiple sets of video and text related data in a one-to-one correspondence; The step of screening and obtaining the public opinion video text combination data corresponding to the preset public opinion topic in the public database includes:

   In the video text combination sub-database, the public opinion video text combination data is filtered according to the keywords of the preset public opinion topic, wherein each of the public opinion video text combination data consists of one public opinion video sub-data and one public opinion Corresponding association composition of text sub-data;

   According to the text similarity, the public opinion text data whose similarity with the text data of the public opinion video text combined data is higher than a first preset value is filtered in the public database; and according to the video source address and the video similarity The step of screening the public opinion video data with the video data of the public opinion video text combination data whose similarity is higher than a second preset value in the public database includes:

   In the separate text sub-database, the public opinion text data whose similarity with the public opinion text sub-data is higher than the first preset value is filtered according to the text similarity, and in the separate video sub-database, according to The video source address and the video similarity are screened to obtain the public opinion video data whose similarity with the public opinion video sub-data is higher than the second preset value.
3. The method for tracking public opinion of video text combination data according to claim 2, wherein said public opinion video text combination is obtained by filtering according to keywords of said preset public opinion topic in said video text combination sub-database The data steps include:

   Parse the preset public opinion topic to obtain at least one first public opinion keyword;

   In the video text combination subdatabase, at least one group of video text combination subdata containing the first public opinion keyword is obtained by filtering, and the video text combination subdata is composed of text subdata and video subdata correspondingly;

   Respectively calculating the number of appearances of each of the video sub-data;

   Retrieve the preset times, compare each of the appearance times with the preset times, and select the video sub-data corresponding to the appearance times greater than the preset times as the public opinion video sub-data;

   From the video text combination sub-database, respectively filter the text sub-data corresponding to each of the public opinion video sub-data as the public opinion text sub-data;

   Each of the public opinion text sub-data and each of the public opinion video sub-data are respectively one-to-one corresponding to obtain the public opinion video text combined data.
4. The method for tracking public opinion of video text combined data according to claim 3, wherein said public opinion text sub-data and each said public opinion video sub-data are respectively corresponded one-to-one to obtain said public opinion video text combination The data steps include:

   Obtain the public address of each public opinion text sub-data and the public address of each public opinion video sub-data;

   Grouping the public opinion text sub-data and the public opinion video sub-data with the same public address in association with each other, so as to achieve a one-to-one correspondence between each of the public opinion text sub-data and each of the public opinion video sub-data;

   The public opinion video text combined data is obtained according to each of the public opinion text sub-data and each of the public opinion video sub-data after being associated and grouped.
5. The public opinion tracking method of video text combined data according to claim 2, wherein the separate video sub-database includes multiple sets of separate video data, and the separate video sub-database is based on the video source address and the video The step of similarity screening to obtain public opinion video data whose similarity with the public opinion video sub-data is higher than the second preset value includes:

   Obtain the original source addresses of each of the individual video data and the public opinion video sub-data;

   Compare the original source address of each of the individual video data with the original source address of the public opinion video sub-data one by one, select the individual video data with the same original source address as the first video data, and select those with inconsistent original source addresses Separate video data as the second video data;

   Respectively calculating video similarity values between each of the second video data and the public opinion video sub-data;

   Respectively comparing each of the video similarity values with the second preset value one by one, and selecting second video data corresponding to a video similarity value greater than the second preset value as the third video data;

   Use the first video data and the third video data as public opinion video data.
6. The method for tracking public opinion of video text combined data according to claim 2, wherein in the separate text sub-database, the similarity with the public opinion text sub-data is higher than that obtained by filtering according to text similarity. The step of the public opinion text data of the first preset value includes:

   According to the part of speech, each of the public opinion text sub-data and each of the individual text data are respectively analyzed to obtain a preset number of second public opinion keywords corresponding to each of the public opinion text sub-data, and all the individual text data corresponding to each. Separate keywords describing the preset number;

   Compare each of the second public opinion keywords with each of the individual keywords, and filter to obtain the same keywords;

   Respectively count the number of appearances corresponding to each of the same keywords;

   Respectively comparing each of the occurrence times with a preset first threshold, and selecting the same keywords corresponding to the occurrence times greater than the first threshold as the designated keywords;

   Select individual text data containing the specified keywords as public opinion text data.
7. The public opinion tracking method of video-text combined data according to claim 1, wherein the designated public data includes the public time of the information, the public opinion system is pre-associated with at least one preset terminal, and the public opinion is based on the public opinion. Data, after the step of obtaining the trend of change in the popularity of the preset public opinion topic includes:

   Judging whether the trend of heat change meets the conditions for triggering automatic push;

   If the conditions for triggering automatic push are met, the designated push public opinion data is filtered from the public opinion data, and the push public opinion data is the data with the shortest time of publication from the current time;

   Acquiring the link of the pushed public opinion data, and generating push information including the link of the pushed public opinion data;

   Sending the push information to each of the preset terminals respectively.
8. A public opinion tracking device for video text combined data is characterized in that it comprises:

   The first obtaining module is configured to obtain designated public data from the first preset network platform according to the first preset frequency;

   A building module for building a public database based on the designated public data;

   The first screening module is configured to filter the public opinion video text combination data corresponding to the preset public opinion topic in the public database;

   The second filtering module is configured to filter the public opinion text data with the text data of the public opinion video text combined data that is higher than the first preset value in the public database according to the text similarity; and according to the video source Address and video similarity, filtering the public opinion video data with the video data of the public opinion video text combination data that has a similarity higher than a second preset value from the public database;

   The first generating module is configured to obtain public opinion data according to the public opinion video text combination data, the public opinion text data, and the public opinion video data;

   The analysis module is used to obtain the trend of the hotness change of the preset public opinion topic according to the public opinion data.
9. The public opinion tracking device for video text combined data according to claim 8, wherein the first screening module comprises:

   The first screening sub-module is configured to obtain the public opinion video text combination data by filtering according to the keywords of the preset public opinion topic in the video text combination sub-database;

   The second screening module includes:

   The second screening sub-module is configured to filter in the separate text sub-database according to the text similarity to obtain public opinion text data whose similarity with the public opinion text sub-data is higher than the first preset value;

   The third screening sub-module is used to filter and obtain public opinion video data whose similarity with the public opinion video sub-data is higher than the second preset value in the separate video sub-database according to the video source address and video similarity .
10. The public opinion tracking device for video text combined data according to claim 9, wherein the first screening submodule comprises:

    The first analysis unit is configured to analyze the preset public opinion topic to obtain at least one first public opinion keyword;

    The first screening unit is configured to filter in the video text combination sub-database to obtain at least one group of video text combination data containing the first public opinion keyword as the video text combination sub-data;

    The first calculation unit is configured to calculate the number of appearances of each of the video sub-data;

    The first selection unit is configured to retrieve a preset number of times, and respectively compare each of the number of appearances with the preset number of times, and select the video sub-data corresponding to the number of appearances greater than the preset number as public opinion Video sub-data;

    The second filtering unit is configured to filter the text sub-data corresponding to each of the public opinion video sub-data as the public opinion text sub-data from the video text combination sub-database;

    The corresponding unit is used to respectively correspond each of the public opinion text sub-data with each of the public opinion video sub-data to obtain the public opinion video text combined data.
11. The public opinion tracking device for video text combined data according to claim 10, wherein the corresponding unit comprises:

    The obtaining subunit is used to obtain the public address of each public opinion text sub-data and the public address of each public opinion video sub-data;

    The association subunit is used to associate and group the public opinion text subdata and the public opinion video subdata with the same public address, so as to realize one-to-one correspondence between each public opinion text subdata and each public opinion video subdata ；

    The generating subunit is configured to obtain the public opinion video text combined data according to each of the public opinion text sub-data and each of the public opinion video sub-data after being associated and grouped.
12. The public opinion tracking device for video text combined data according to claim 9, wherein the third screening sub-module comprises:

    An obtaining unit, configured to obtain the original source addresses of each of the individual video data and the public opinion video sub-data;

    The second comparison unit is configured to compare the original source addresses of the individual video data with the original source addresses of the public opinion video sub-data one by one, and select the individual video data with the same original source addresses as the first video data, And select separate video data with inconsistent original source addresses as the second video data;

    A second calculation unit, configured to calculate video similarity values between each of the second video data and the public opinion video sub-data;

    The second selection unit is configured to retrieve a second preset value, compare each of the video similarity values with the second preset value one by one, and select a video similarity value greater than the second preset value The corresponding second video data is used as the third video data;

    The setting unit is configured to use the first video data and the third video data as public opinion video data.
13. The public opinion tracking device for combined video text data according to claim 9, wherein the second screening submodule further comprises:

    The second parsing unit is used to analyze each of the public opinion text sub-data and each of the individual text data according to the part of speech to obtain a preset number of second public opinion keywords corresponding to each of the public opinion text sub-data, and each post The preset number of individual keywords corresponding to the individual text data;

    The comparison unit is used to compare each of the second public opinion keywords with each of the individual keywords, and obtain the same keywords through screening;

    The statistical unit is used to separately count the number of appearances corresponding to each of the same keywords;

    The third selection unit is configured to compare each of the occurrence times with a preset first threshold, and select the same keywords corresponding to the occurrence times greater than the first threshold as designated keywords;

    The fourth selection unit is used to select individual text data containing the designated keywords as public opinion text data.
14. The public opinion tracking device for video text combined data according to claim 8, wherein the public opinion tracking device further comprises:

    A judging module for judging whether the heat change trend meets the conditions for triggering automatic push;

    The third screening module is used to screen the public opinion data corresponding to the public opinion with the shortest distance to the current time from the public opinion data as the push public opinion data;

    The second generating module is used to obtain the link of the pushed public opinion data and generate push information including the link of the pushed public opinion data;

    The sending module is configured to send the push information to each of the preset terminals respectively.
15. A computer device includes a memory and a processor, wherein computer-readable instructions are stored in the memory, and is characterized in that when the processor is executed, a public opinion tracking method for video text combined data is realized, and the public opinion tracking method includes:

    According to a first preset frequency, obtain designated public data from a first preset network platform, the designated public data including all the public information of the first preset network platform and the propagation path corresponding to each public information , The public information includes individual text information with only text, individual video information with only video, and combined video and text information related to the video and the text;

    Construct a public database according to the specified public data;

    Screening and obtaining public opinion video text combination data corresponding to the preset public opinion topic in the public database;

    According to the text similarity, the public opinion text data whose similarity with the text data of the public opinion video text combined data is higher than the first preset value is filtered in the public database; and according to the video source address and the video similarity, The public opinion video data whose similarity with the video data of the public opinion video text combined data is higher than a second preset value is filtered from the public database;

    Obtaining public opinion data according to the public opinion video text combination data, the public opinion text data, and the public opinion video data;

    According to the public opinion data, the popularity trend of the preset public opinion topic is obtained.
16. The computer device according to claim 15, wherein the public database includes a separate text sub-database, a separate video sub-database, and a video text combined sub-database, and the separate text sub-database is composed of multiple sets of separate text data A database, the separate video sub-database is a database composed of multiple sets of separate video data, the video-text combination sub-database is a database composed of multiple sets of video and text associated data in a one-to-one correspondence; the processor executes in the The step of screening and obtaining the public opinion video text combination data corresponding to the preset public opinion topic in the public database includes:

    In the video text combination sub-database, the public opinion video text combination data is filtered according to the keywords of the preset public opinion topic, wherein each of the public opinion video text combination data consists of one public opinion video sub-data and one public opinion Corresponding association composition of text sub-data;

    According to the text similarity, the public opinion text data whose similarity with the text data of the public opinion video text combined data is higher than a first preset value is filtered in the public database; and according to the video source address and the video similarity The step of screening the public opinion video data with the video data of the public opinion video text combination data whose similarity is higher than a second preset value in the public database includes:

    In the separate text sub-database, the public opinion text data whose similarity with the public opinion text sub-data is higher than the first preset value is filtered according to the text similarity, and in the separate video sub-database, according to The video source address and the video similarity are screened to obtain the public opinion video data whose similarity with the public opinion video sub-data is higher than the second preset value.
17. The computer device according to claim 16, wherein the processor executes the step of obtaining the public opinion video text combination data in the video text combination sub-database, filtering according to the keywords of the preset public opinion topic ,include:

    Parse the preset public opinion topic to obtain at least one first public opinion keyword;

    In the video text combination subdatabase, at least one group of video text combination subdata containing the first public opinion keyword is obtained by filtering, and the video text combination subdata is composed of text subdata and video subdata correspondingly;

    Respectively calculating the number of appearances of each of the video sub-data;

    Retrieve the preset times, compare each of the appearance times with the preset times, and select the video sub-data corresponding to the appearance times greater than the preset times as the public opinion video sub-data;

    From the video text combination sub-database, respectively filter the text sub-data corresponding to each of the public opinion video sub-data as the public opinion text sub-data;

    Each of the public opinion text sub-data and each of the public opinion video sub-data are respectively one-to-one corresponding to obtain the public opinion video text combined data.
18. A computer non-volatile readable storage medium having computer readable instructions stored thereon, wherein the computer readable instructions are executed by a processor to implement a public opinion tracking method for video text combined data, and the public opinion tracking method include:

    According to a first preset frequency, obtain designated public data from a first preset network platform, the designated public data including all the public information of the first preset network platform and the propagation path corresponding to each public information , The public information includes individual text information with only text, individual video information with only video, and combined video and text information related to the video and the text;

    Construct a public database according to the specified public data;

    Screening and obtaining public opinion video text combination data corresponding to the preset public opinion topic in the public database;

    According to the text similarity, the public opinion text data whose similarity with the text data of the public opinion video text combined data is higher than the first preset value is filtered in the public database; and according to the video source address and the video similarity, The public opinion video data whose similarity with the video data of the public opinion video text combined data is higher than a second preset value is filtered from the public database;

    Obtaining public opinion data according to the public opinion video text combination data, the public opinion text data, and the public opinion video data;

    According to the public opinion data, the popularity trend of the preset public opinion topic is obtained.
19. The computer non-volatile readable storage medium according to claim 18, wherein the public database includes a separate text sub-database, a separate video sub-database, and a video text combined sub-database, and the separate text sub-database is multiple A database composed of separate sets of text data, the separate video sub-database is a database composed of multiple sets of separate video data, and the video-text combined sub-database is a database composed of multiple sets of video and text related data in a one-to-one correspondence; The processor executes the step of screening the public opinion video text combination data corresponding to the preset public opinion topic in the public database, including:

    In the video text combination sub-database, the public opinion video text combination data is filtered according to the keywords of the preset public opinion topic, wherein each of the public opinion video text combination data consists of one public opinion video sub-data and one public opinion Corresponding association composition of text sub-data;

    According to the text similarity, the public opinion text data whose similarity with the text data of the public opinion video text combined data is higher than a first preset value is filtered in the public database; and according to the video source address and the video similarity The step of screening the public opinion video data with the video data of the public opinion video text combination data whose similarity is higher than a second preset value in the public database includes:

    In the separate text sub-database, the public opinion text data whose similarity with the public opinion text sub-data is higher than the first preset value is filtered according to the text similarity, and in the separate video sub-database, according to The video source address and the video similarity are screened to obtain the public opinion video data whose similarity with the public opinion video sub-data is higher than the second preset value.
20. The computer non-volatile readable storage medium according to claim 19, wherein the processor executes in the video text combination sub-database to filter according to the keywords of the preset public opinion topic to obtain the The steps of combining public opinion video text data include:

    Parse the preset public opinion topic to obtain at least one first public opinion keyword;

    In the video text combination subdatabase, at least one group of video text combination subdata containing the first public opinion keyword is obtained by filtering, and the video text combination subdata is composed of text subdata and video subdata correspondingly;

    Respectively calculating the number of appearances of each of the video sub-data;

    Retrieve the preset times, compare each of the appearance times with the preset times, and select the video sub-data corresponding to the appearance times greater than the preset times as the public opinion video sub-data;

    From the video text combination sub-database, respectively filter the text sub-data corresponding to each of the public opinion video sub-data as the public opinion text sub-data;

    Each of the public opinion text sub-data and each of the public opinion video sub-data are respectively one-to-one corresponding to obtain the public opinion video text combined data.