Deep learning-based traditional culture digitization international communication effect evaluation method

Abstract

The invention relates to the technical field of cultural transmission effect recognition, in particular to a traditional culture digitization international transmission effect evaluation method based on deep learning. Obtaining user data of the international network social platform in a first period, and constructing to obtain a first data set; performing retrieval according to the retrieval keyword to obtain first cultural social data and a first user; constructing a culture interest recognition model to recognize the first culture social data to obtain a first culture interest coefficient; identifying the user position and the first cultural interest coefficient according to a clustering algorithm to obtain first distribution data; obtaining user data of the international network social platform in a second period, constructing to obtain a second data set, and identifying to obtain a second user, a second cultural interest coefficient and second distribution data; and obtaining regional distribution data according to the first distribution data and the second distribution data, and measuring and calculating to obtain a cultural transmission effect coefficient. According to the method, the transmission effect of the traditional culture is accurately measured through the culture transmission effect coefficient.

Description

Technical Field

[0001] The present invention relates to the technical field of cultural communication effect recognition, and specifically to a method for evaluating the international communication effect of digital traditional culture based on deep learning. Background Art

[0002] With the rapid development of Internet technology, social platforms have become an important channel for modern information dissemination, promoting instant communication and interactive exchanges. In particular, international social platforms have profoundly changed the way global information flows, expanded the boundaries of cultural dissemination, and promoted cross-cultural exchanges and the integration of multiple cultures.

[0003] The dissemination of traditional culture is not only the protection of historical heritage, but also the strengthening of modern social cultural identity and respect for global cultural diversity. As globalization deepens, cultural exchanges and collisions are becoming more frequent, and the dissemination of traditional culture is particularly important. However, in the process of traditional culture dissemination, it is often restricted by multiple factors such as language, region, and concepts, resulting in limited dissemination effects.

[0004] With the global development of social platforms, these limitations have been gradually broken, allowing people from different cultural backgrounds to communicate and learn in an interactive and open environment. With the sharing and interactive functions of the platform, traditional culture can be spread in a more vivid, interesting and interactive way, covering a wider global audience.

[0005] At present, the existing communication effect evaluation methods are mostly concentrated in the traditional media or marketing fields, lacking a comprehensive evaluation framework for the characteristics of cultural communication, especially at the international communication level. Therefore, it is urgent to develop a more comprehensive and accurate evaluation method to measure the effect of digital traditional cultural communication and improve the effectiveness of international communication strategies.

[0006] To this end, a method for evaluating the effectiveness of international communication of digital traditional culture based on deep learning is proposed. Summary of the invention

[0007] The purpose of the present invention is to provide a method for evaluating the effect of international communication of digital traditional culture based on deep learning, by acquiring user data of an international network social platform in a first period, constructing a first data set; retrieving first cultural social data and a first user according to a search index; constructing a cultural interest recognition model to recognize the first cultural social data, and obtaining a first cultural interest coefficient; recognizing the user position and the first cultural interest coefficient according to a clustering algorithm, and obtaining first distribution data; acquiring user data of an international network social platform in a second period, constructing a second data set, and identifying a second user, a second cultural interest coefficient, and a second distribution data; obtaining regional distribution data according to the first distribution data and the second distribution data, and calculating a cultural communication effect coefficient. The present invention accurately measures the communication effect of traditional culture through the cultural communication effect coefficient.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] The evaluation method of the international communication effect of traditional culture digitization based on deep learning includes:

[0010] S10. Obtain user data of the international network social platform in the first period to construct a first data set; the user data includes user age, user location and user social data; the user social data includes browsing comment data and posting interaction data;

[0011] S20. Determine a search index according to the type of traditional culture, and search the user social data in the first data set according to the search index to obtain the first cultural social data and the corresponding first user;

[0012] S30. Constructing a cultural interest recognition model to recognize the first cultural social data, and obtaining a first cultural interest coefficient according to the first browsing comment data and the first posting interaction data in the first cultural social data;

[0013] S40. Identify the user location and the first cultural interest coefficient of the first user according to the clustering algorithm to obtain first distribution data of the first user in the geographical dimension;

[0014] S50. Obtain user data of the international network social platform in the second period to construct a second data set; identify the second data set according to the method of retrieving and identifying the first user, the first cultural interest coefficient and the first distribution data of the first data set to obtain the second user, the second cultural interest coefficient and the second distribution data;

[0015] S60. Obtain regional distribution data based on the first distribution data and the second distribution data; the regional distribution data includes a first distribution area, a second distribution area and a third distribution area; and obtain a cultural communication effect coefficient based on the user age, cultural interest coefficient, user density and regional area in the regional distribution data.

[0016] The cultural interest recognition model is constructed based on deep learning, and includes a browsing behavior recognition layer, a posting behavior recognition layer, and a cultural interest recognition layer;

[0017] The browsing behavior recognition layer divides the browsing comment data according to the browsing content to obtain video browsing comment data and picture and text browsing comment data; obtains a first browsing coefficient according to the video browsing comment data; obtains a second browsing coefficient according to the picture and text browsing comment data; and obtains a cultural browsing coefficient according to the first browsing coefficient and the second browsing coefficient.

[0018] The posting behavior recognition layer divides the posting interaction data according to the posting content to obtain video posting interaction data and picture and text posting interaction data; obtains a first posting coefficient according to the video posting interaction data; obtains a second posting coefficient according to the picture and text posting interaction data; and obtains a cultural posting coefficient according to the first posting coefficient and the second posting coefficient.

[0019] The cultural interest identification layer calculates the cultural interest coefficient based on the cultural browsing coefficient and the cultural posting coefficient.

[0020] The browsing behavior recognition layer recognizes video browsing comment data and picture and text browsing comment data in the following process:

[0021] According to the video browsing comment data, the video browsing time of the user and the total time of the video, the first video data is obtained; according to the video browsing comment data, the number of video comments, the video comment vocabulary and the video comment sentiment coefficient of the user, the second video data is obtained; according to the first video data and the second video data, the first browsing coefficient is calculated;

[0022] The first picture and text data is obtained based on the user's picture and text browsing time and the platform average browsing time of the pictures and texts in the picture and text browsing and commenting data; the second picture and text data is obtained based on the number of picture and text comments, the picture and text comment vocabulary and the picture and text comment sentiment coefficient in the picture and text browsing and commenting data; the second browsing coefficient is calculated based on the first picture and text data and the second picture and text data.

[0023] The posting behavior recognition layer recognizes video posting interaction data and picture and text posting interaction data in the following process:

[0024] According to the video posting interaction data, the image content, audio content and subtitle content of the video are identified to obtain the video emotion coefficient; according to the video posting interaction data, the video interaction times, video interaction vocabulary and video interaction emotion coefficient of the posting user are obtained to obtain the first interaction data; according to the video emotion coefficient, the posting video duration and the first interaction data, the first posting coefficient is obtained;

[0025] According to the interactive data of picture and text posting, the image content and text content of the picture and text are identified to obtain the picture and text sentiment coefficient; according to the picture and text interaction times, the picture and text interaction vocabulary and the picture and text interaction sentiment coefficient of the posting user in the picture and text posting interactive data, the second interactive data is obtained; according to the picture and text sentiment coefficient, the number of posted pictures and texts and the second interactive data, the second posting coefficient is obtained.

[0026] The process of identifying and dividing user locations and first cultural interest coefficients through clustering algorithms is as follows:

[0027] A clustering algorithm based on density dimension is selected to identify user locations and first cultural interest coefficients, and the first cultural interest coefficient is used as the weight of the user location to act on distance correction and density identification between user locations;

[0028] User locations whose user location density is greater than a preset density threshold are divided to obtain first distribution data.

[0029] The process of obtaining the regional distribution data is as follows:

[0030] Overlap the geographical dimensions of the first distribution data and the second distribution data accordingly; take the overlapping area of ​​the first distribution data and the second distribution data as the first distribution area; take the area of ​​the first distribution data that does not overlap with the second distribution data as the second distribution area; take the area of ​​the second distribution data that does not overlap with the first distribution data as the third distribution area.

[0031] According to the average age change between the first user and the second user, the cultural interest coefficient change, the user density change and the area of ​​the first region within the first distribution area, a first propagation coefficient is obtained;

[0032] The second propagation coefficient is calculated based on the average age of the first user, the first cultural interest coefficient, the user density, and the area of ​​the second region within the second distribution area;

[0033] The third propagation coefficient is calculated based on the average age of the second users, the second cultural interest coefficient, the user density and the area of ​​the third region within the third distribution area;

[0034] The cultural communication effect coefficient is calculated based on the first communication coefficient, the second communication coefficient and the third communication coefficient.

[0035] The formula of the cultural communication effect coefficient is:

[0036] CulF=α1*Eff1+α2*Eff2+α3*Eff3;

[0037] Among them, CulF represents the cultural communication effect coefficient; α1 represents the first communication weight; Eff1 represents the first communication coefficient; α2 represents the second communication weight; Eff2 represents the second communication coefficient; α3 represents the third communication weight; Eff3 represents the third communication coefficient;

[0038]

[0039] Among them, area1 represents the area of ​​the first distribution area; β1 represents the first age weight; represents the average age of the first users in the first distribution area in the first period; represents the average age of the second users in the first distribution area in the second period; AD represents the time difference between the second period and the first period; AT represents the age change threshold; γ1 represents the first cultural interest weight; represents the average cultural interest coefficient of the second users in the first distribution area in the second period; represents the average value of the first user's cultural interest coefficient in the first distribution area in the first period; CT represents the cultural interest coefficient threshold; η1 represents the first density weight; represents a second user density of the first distribution area in a second period; represents the first user density of the first distribution area in the first period; DT represents the user density threshold;

[0040]

[0041] Among them, area2 represents the area of ​​the second distribution area; β2 represents the second age weight; represents the average age of the first users in the second distribution area in the first period; γ2 represents the second cultural interest weight; represents the average value of the first user's cultural interest coefficient in the second distribution area in the first period; η2 represents the second density weight; represents a first user density of a second distribution area in a first period;

[0042]

[0043] Among them, area3 represents the area of ​​the third distribution area; β3 represents the third age weight; represents the average age of the second users in the third distribution area in the second period; γ3 represents the third cultural interest weight; represents the average cultural interest coefficient of the second user in the third distribution area in the second period; η3 represents the third density weight; It represents the second user density of the third distribution area in the second period.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] 1. The present invention constructs a cultural interest recognition model based on deep learning; divides the browsing comment data through the model to obtain video browsing comment data and picture and text browsing comment data; obtains a first browsing coefficient based on the video browsing comment data; obtains a second browsing coefficient based on the picture and text browsing comment data; obtains a cultural browsing coefficient based on the first browsing coefficient and the second browsing coefficient; then divides the posting interaction data through the model to obtain video posting interaction data and picture and text posting interaction data; obtains a first posting coefficient based on the video posting interaction data; obtains a second posting coefficient based on the picture and text posting interaction data; obtains a cultural posting coefficient based on the first posting coefficient and the second posting coefficient; finally, according to the cultural browsing coefficient and the cultural posting coefficient, the user's cultural interest coefficient in traditional culture is accurately calculated.

[0046] 2. The present invention uses a clustering algorithm based on the measurement dimension of user density to identify user locations and cultural interest coefficients, and uses the cultural interest coefficient as the weight of the user location to act on the distance correction and density identification between user locations; then the user locations whose user location density is greater than a preset density threshold are classified into one category to obtain distribution data; the density distribution of users is accurately identified and divided to obtain regional distribution data of users interested in traditional culture.

[0047] 3. The present invention obtains first distribution data of a first period and second distribution data of a second period; overlaps the geographical dimensions of the first distribution data and the second distribution data accordingly; obtains regional distribution data based on the overlapping distribution of the first distribution data and the second distribution data; accurately identifies the distribution changes of users between the first period and the second period; and then obtains the cultural communication effect coefficient based on the user age, cultural interest coefficient, user density and region in the regional distribution data, so as to accurately measure the cultural communication effect of the second period relative to the first period. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a flow chart of the method for evaluating the effect of international communication of digital traditional culture based on deep learning of the present invention;

[0049] Figure 2 It is a structural schematic diagram of the cultural interest identification model of the present invention;

[0050] Figure 3This is a schematic diagram of obtaining regional distribution data of the present invention.

[0051] In the figure: 10 represents the first distribution data; 20 represents the second distribution data; 30 represents the regional distribution data; 31 represents the first distribution region; 32 represents the second distribution region; 33 represents the third distribution region. DETAILED DESCRIPTION

[0052] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] Embodiment 1

[0054] The present invention proposes a method for evaluating the effect of international communication of traditional culture digitization based on deep learning, and its process is as follows: Figure 1 Shown; including:

[0055] S10. Obtain user data of an international online social platform in a first period to construct a first data set; the user data includes user age, user location and user social data; the user social data includes browsing comment data and posting interaction data.

[0056] The browsing and commenting data includes browsing and commenting data of users on international social networking platforms; the commenting data includes the number of comments, commenting vocabulary, and evaluation operations; the evaluation operations include operations such as liking, blocking, and recommending posts by other users;

[0057] The posting interaction data includes the user's posting data on the international network social platform and the interaction data with other users under the posting data, such as interactive comments, likes data and sharing data.

[0058] S20. Determine a search index according to the type of traditional culture, and search the user social data in the first data set according to the search index to obtain the first cultural social data and the corresponding first user.

[0059] The process of obtaining the search index is as follows: determining the type of traditional culture; establishing an expert group according to the type of traditional culture; selecting experts in the field of traditional culture and experts in the field of network communication to determine the search index of the traditional culture on the international network social platform; the search index includes search keywords, search symbols, search images, search audio and search video, etc.

[0060] S30. Construct a cultural interest recognition model to recognize the first cultural social data, and obtain a first cultural interest coefficient according to the first browsing comment data and the first posting interaction data in the first cultural social data.

[0061] The cultural interest recognition model is constructed based on deep learning, and its structure is as follows: Figure 2 As shown, it includes browsing behavior recognition layer, posting behavior recognition layer and cultural interest recognition layer;

[0062] The browsing behavior recognition layer divides the browsing comment data according to the browsing content to obtain video browsing comment data and picture and text browsing comment data; obtains a first browsing coefficient according to the video browsing comment data; obtains a second browsing coefficient according to the picture and text browsing comment data; and obtains a cultural browsing coefficient according to the first browsing coefficient and the second browsing coefficient.

[0063] The posting behavior recognition layer divides the posting interaction data according to the posting content to obtain video posting interaction data and picture and text posting interaction data; obtains a first posting coefficient according to the video posting interaction data; obtains a second posting coefficient according to the picture and text posting interaction data; and obtains a cultural posting coefficient according to the first posting coefficient and the second posting coefficient.

[0064] The cultural interest identification layer calculates the cultural interest coefficient based on the cultural browsing coefficient and the cultural posting coefficient.

[0065] The present invention obtains a cultural interest recognition model based on deep learning construction; the browsing comment data is divided by the model to obtain video browsing comment data and picture and text browsing comment data; a first browsing coefficient is obtained according to the video browsing comment data identification; a second browsing coefficient is obtained according to the picture and text browsing comment data identification; a cultural browsing coefficient is obtained by measuring the first browsing coefficient and the second browsing coefficient; the posting interaction data is then divided by the model to obtain video posting interaction data and picture and text posting interaction data; a first posting coefficient is obtained according to the video posting interaction data identification; a second posting coefficient is obtained according to the picture and text posting interaction data identification; a cultural posting coefficient is obtained by measuring the first posting coefficient and the second posting coefficient; finally, according to the cultural browsing coefficient and the cultural posting coefficient, the user's cultural interest coefficient for traditional culture is accurately calculated.

[0066] The browsing behavior recognition layer recognizes video browsing comment data and picture and text browsing comment data in the following process:

[0067] According to the video browsing comment data, the video browsing time of the user and the total time of the video, the first video data is obtained; according to the video browsing comment data, the number of video comments, the video comment vocabulary and the video comment sentiment coefficient of the user, the second video data is obtained; according to the first video data and the second video data, the first browsing coefficient is calculated;

[0068] The first picture and text data is obtained based on the user's picture and text browsing time and the platform average browsing time of the pictures and texts in the picture and text browsing and commenting data; the second picture and text data is obtained based on the number of picture and text comments, the picture and text comment vocabulary and the picture and text comment sentiment coefficient in the picture and text browsing and commenting data; the second browsing coefficient is calculated based on the first picture and text data and the second picture and text data.

[0069] The present invention obtains first video data according to the user's video browsing time and the total time of the video; obtains second video data according to the user's video comment times, video comment vocabulary and video comment sentiment coefficient; obtains a first browsing coefficient by calculation according to the first video data and the second video data; obtains first picture and text data according to the user's picture and text browsing time and the platform average browsing time of the pictures and texts; obtains second picture and text data according to the user's picture and text comment times, picture and text comment vocabulary and picture and text comment sentiment coefficient; obtains a second browsing coefficient by calculation according to the first picture and text data and the second picture and text data; and accurately identifies the degree of bias of the user in the process of browsing traditional cultural related content.

[0070] The posting behavior recognition layer recognizes video posting interaction data and picture and text posting interaction data in the following process:

[0071] According to the video posting interaction data, the image content, audio content and subtitle content of the video are identified to obtain the video emotion coefficient; according to the video posting interaction data, the video interaction times, video interaction vocabulary and video interaction emotion coefficient of the posting user are obtained to obtain the first interaction data; according to the video emotion coefficient, the posting video duration and the first interaction data, the first posting coefficient is obtained;

[0072] According to the interactive data of picture and text posting, the image content and text content of the picture and text are identified to obtain the picture and text sentiment coefficient; according to the picture and text interaction times, the picture and text interaction vocabulary and the picture and text interaction sentiment coefficient of the posting user in the picture and text posting interactive data, the second interactive data is obtained; according to the picture and text sentiment coefficient, the number of posted pictures and texts and the second interactive data, the second posting coefficient is obtained.

[0073] The present invention identifies the image content, audio content and subtitle content of a video in video posting interaction data to obtain a video emotion coefficient; obtains first interaction data according to the video interaction times, video interaction vocabulary and video interaction emotion coefficient of a posting user; obtains a first posting coefficient according to the video emotion coefficient, posting video duration and the first interaction data; identifies the image content and text content of a picture and text in picture and text posting interaction data to obtain a picture and text emotion coefficient; obtains second interaction data according to the picture and text interaction times, picture and text interaction vocabulary and picture and text interaction emotion coefficient of the posting user; obtains a second posting coefficient according to the picture and text emotion coefficient, the number of posted pictures and texts and the second interaction data; and accurately identifies the degree of preference for traditional culture in the posting process of the user.

[0074] S40. Identify the user location and the first cultural interest coefficient of the first user according to a clustering algorithm to obtain first distribution data of the first user in a geographical dimension.

[0075] The process of identifying and dividing user locations and first cultural interest coefficients through clustering algorithms is as follows:

[0076] A clustering algorithm based on density dimension is selected to identify user locations and first cultural interest coefficients, and the first cultural interest coefficient is used as the weight of the user location to act on distance correction and density identification between user locations;

[0077] User locations whose user location density is greater than a preset density threshold are classified into one category to obtain first distribution data.

[0078] In the process of spreading traditional culture, the spreading effect of the traditional culture interest area is better than that of the scattered interested individuals; the present invention identifies the distribution area where users gather according to the user density, and effectively improves the recognition accuracy of the spreading effect.

[0079] The present invention identifies user locations and cultural interest coefficients based on the measurement dimension of user density through a clustering algorithm, uses the cultural interest coefficient as the weight of the user location, and acts on the distance correction and density identification between user locations; then divides the user locations whose user location density is greater than a preset density threshold into one category to obtain distribution data; accurately identifies and divides the user's density distribution to obtain distribution data of users interested in traditional culture.

[0080] S50. Obtain user data of the international network social platform in the second period to construct a second data set; identify the second data set according to the method of retrieving and identifying the first user, the first cultural interest coefficient and the first distribution data of the first data set to obtain the second user, the second cultural interest coefficient and the second distribution data.

[0081] S60. Obtain regional distribution data based on the first distribution data and the second distribution data; the regional distribution data includes a first distribution area, a second distribution area and a third distribution area; and obtain a cultural communication effect coefficient based on the user age, cultural interest coefficient, user density and regional area in the regional distribution data.

[0082] The process of obtaining the regional distribution data is as follows:

[0083] Overlap the geographical dimensions of the first distribution data and the second distribution data accordingly; take the overlapping area of ​​the first distribution data and the second distribution data as the first distribution area; take the area of ​​the first distribution data that does not overlap with the second distribution data as the second distribution area; take the area of ​​the second distribution data that does not overlap with the first distribution data as the third distribution area.

[0084] The present invention obtains first distribution data of a first period and second distribution data of a second period; overlaps the geographical dimensions of the first distribution data and the second distribution data accordingly; obtains regional distribution data based on the overlapping distribution of the first distribution data and the second distribution data; and accurately identifies the distribution changes of users between the first period and the second period.

[0085] The first propagation coefficient is obtained according to the average age change, cultural interest coefficient change, user density change and the area of ​​the first area between the first user and the second user in the first distribution area; the second propagation coefficient is obtained by measuring the average age, the first cultural interest coefficient, the user density and the area of ​​the second area of ​​the first user in the second distribution area; the third propagation coefficient is obtained by measuring the average age, the second cultural interest coefficient, the user density and the area of ​​the third area of ​​the second user in the third distribution area;

[0086] The formula of the cultural communication effect coefficient is:

[0087] CulF=α1*Eff1+α2*Eff2+α3*Eff3;

[0088] Among them, CulF represents the cultural communication effect coefficient; α1 represents the first communication weight; Eff1 represents the first communication coefficient; α2 represents the second communication weight, which can be negative; Eff2 represents the second communication coefficient; α3 represents the third communication weight; Eff3 represents the third communication coefficient;

[0089]

[0090] Among them, area1 represents the area of ​​the first distribution area; β1 represents the first age weight; represents the average age of the first users in the first distribution area in the first period; represents the average age of the second users in the first distribution area in the second period; AD represents the time difference between the second period and the first period; AT represents the age change threshold; γ1 represents the first cultural interest weight; represents the average cultural interest coefficient of the second users in the first distribution area in the second period; represents the average value of the first user's cultural interest coefficient in the first distribution area in the first period; CT represents the cultural interest coefficient threshold; η1 represents the first density weight; represents a second user density of the first distribution area in a second period; represents the first user density of the first distribution area in the first period; DT represents the user density threshold;

[0091] The first distribution area is an area where the user density is higher than the density threshold in both the first period and the second period. The present invention starts from the perspective of the change in the average value of the user's cultural interest coefficient and the change in the user density, and comprehensively considers the change in the average age of the users. Relatively speaking, young users have higher cultural communication potential than older users; thereby accurately measuring the traditional cultural communication effect of the first distribution area.

[0092]

[0093] Among them, area2 represents the area of ​​the second distribution area; β2 represents the second age weight; represents the average age of the first users in the second distribution area in the first period; γ2 represents the second cultural interest weight; represents the average value of the first user's cultural interest coefficient in the second distribution area in the first period; η2 represents the second density weight; represents a first user density of a second distribution area in a first period;

[0094]

[0095] Among them, area3 represents the area of ​​the third distribution area; β3 represents the third age weight; represents the average age of the second users in the third distribution area in the second period; γ3 represents the third cultural interest weight; represents the average cultural interest coefficient of the second user in the third distribution area in the second period; η3 represents the third density weight; It represents the second user density of the third distribution area in the second period.

[0096] The age weight, cultural interest weight and density weight in the formula are all determined by the expert group based on actual data, and the value range is 0 to 1.

[0097] The present invention obtains the user age, cultural interest coefficient, user density and area of ​​the first distribution area, the second distribution area and the third distribution area in the first period and the second period; calculates the cultural communication effect coefficient according to the changes in the user age, cultural interest and density from the first period to the second period; and accurately measures the cultural communication effect of the second period relative to the first period.

[0098] The present invention obtains user data of an international network social platform in a first period, constructs a first data set; retrieves first cultural social data and a first user according to a search index; constructs a cultural interest recognition model to recognize the first cultural social data, obtains a first cultural interest coefficient; recognizes the user position and the first cultural interest coefficient according to a clustering algorithm, obtains first distribution data; obtains user data of an international network social platform in a second period, constructs a second data set, recognizes a second user, a second cultural interest coefficient, and a second distribution data; obtains regional distribution data according to the first distribution data and the second distribution data, and calculates a cultural communication effect coefficient. The present invention accurately measures the communication effect of traditional culture through the cultural communication effect coefficient.

[0099] Embodiment 2

[0100] This embodiment selects traditional culture T as the research object, and uses the traditional culture digitization international communication effect evaluation method based on deep learning of the present invention to identify its international communication effect.

[0101] The method for evaluating the effect of international communication of digital traditional culture based on deep learning includes:

[0102] S10. Obtain user data of an international online social platform in a first period to construct a first data set; the user data includes user age, user location and user social data; the user social data includes browsing comment data and posting interaction data.

[0103] In the study of the international communication effect of traditional culture T, one year is used as the identification period, so the first period is the first year and the second period is the second year; the time difference between the first period and the second period is one year.

[0104] The browsing and commenting data include browsing data and commenting data of users on international social networking platforms; wherein the commenting data include the number of comments, commenting vocabulary and evaluation operations; the evaluation operations include operations such as liking, blocking and recommending posts by other users; the posting interaction data include posting data of users on international social networking platforms and interaction data with other users under the posting data, such as interactive comments, liked data and shared data.

[0105] S20. Determine a search index according to the type of traditional culture, and search the user social data in the first data set according to the search index to obtain the first cultural social data and the corresponding first user.

[0106] The process of obtaining the search index is as follows: determining the type of traditional culture to be studied, and establishing an expert group according to the type of traditional culture; selecting experts in the traditional culture and experts in the field of network communication, and determining keywords for the traditional culture on international network social platforms.

[0107] S30. Construct a cultural interest recognition model to recognize the first cultural social data, and obtain a first cultural interest coefficient according to the first browsing comment data and the first posting interaction data in the first cultural social data.

[0108] The cultural interest recognition model is constructed based on deep learning, and includes a browsing behavior recognition layer, a posting behavior recognition layer, and a cultural interest recognition layer;

[0109] The browsing behavior recognition layer divides the browsing comment data according to the browsing content to obtain video browsing comment data and picture and text browsing comment data; obtains a first browsing coefficient according to the video browsing comment data; obtains a second browsing coefficient according to the picture and text browsing comment data; and obtains a cultural browsing coefficient according to the first browsing coefficient and the second browsing coefficient.

[0110] The posting behavior recognition layer divides the posting interaction data according to the posting content to obtain video posting interaction data and picture and text posting interaction data; obtains a first posting coefficient according to the video posting interaction data; obtains a second posting coefficient according to the picture and text posting interaction data; and obtains a cultural posting coefficient according to the first posting coefficient and the second posting coefficient.

[0111] The cultural interest identification layer calculates the cultural interest coefficient based on the cultural browsing coefficient and the cultural posting coefficient.

[0112] The present invention constructs a cultural interest recognition model based on deep learning; divides browsing comment data through the model to obtain video browsing comment data and picture and text browsing comment data, and identifies the cultural browsing coefficient; divides posting interaction data through the model to obtain video posting interaction data and picture and text posting interaction data, and identifies the cultural posting coefficient; finally, according to the cultural browsing coefficient and the cultural posting coefficient, the user's cultural interest coefficient for traditional culture is accurately calculated.

[0113] The browsing behavior recognition layer recognizes video browsing comment data and picture and text browsing comment data in the following process:

[0114] According to the video browsing comment data, the video browsing time of the user and the total time of the video, the first video data is obtained; according to the video browsing comment data, the number of video comments, the video comment vocabulary and the video comment sentiment coefficient of the user, the second video data is obtained; according to the first video data and the second video data, the first browsing coefficient is calculated;

[0115] The video comment sentiment coefficient is obtained based on the video comment vocabulary and the behavior recognition in the evaluation operation. The value range of the video comment sentiment coefficient is -1 to 1. When it is less than 0, it indicates negative sentiment. The closer it is to -1, the stronger the negative sentiment is. When it is greater than 0, it indicates positive sentiment. The closer it is to 1, the stronger the positive sentiment is.

[0116] The video browsing and comment data of the first period of traditional culture T are identified, and Table 1 is obtained.

[0117] Table 1 Video browsing comment data table

[0118]

[0119] The first picture and text data is obtained based on the user's picture and text browsing time and the platform average browsing time of the pictures and texts in the picture and text browsing and commenting data; the second picture and text data is obtained based on the number of picture and text comments, the picture and text comment vocabulary and the picture and text comment sentiment coefficient in the picture and text browsing and commenting data; the second browsing coefficient is calculated based on the first picture and text data and the second picture and text data.

[0120] The sentiment coefficient of the picture and text review is obtained based on the vocabulary of the picture and text review and the corresponding evaluation operation behavior. The value range of the sentiment coefficient of the picture and text review is -1 to 1. When it is less than 0, it indicates negative sentiment. The closer it is to -1, the stronger the negative sentiment is. When it is greater than 0, it indicates positive sentiment. The closer it is to 1, the stronger the positive sentiment is.

[0121] The image and text browsing and comment data of the first period of traditional culture T were identified, and Table 2 was obtained.

[0122] Table 2 Picture and text browsing comment data table

[0123]

[0124] The present invention obtains first video data according to the user's video browsing time and the total time of the video; obtains second video data according to the user's video comment times, video comment vocabulary and video comment sentiment coefficient; obtains a first browsing coefficient by calculation according to the first video data and the second video data; obtains first picture and text data according to the user's picture and text browsing time and the platform average browsing time of the pictures and texts; obtains second picture and text data according to the user's picture and text comment times, picture and text comment vocabulary and picture and text comment sentiment coefficient; obtains a second browsing coefficient by calculation according to the first picture and text data and the second picture and text data; and accurately identifies the degree of bias of the user in the process of browsing traditional cultural related content.

[0125] The posting behavior recognition layer recognizes video posting interaction data and picture and text posting interaction data in the following process:

[0126] According to the video posting interaction data, the image content, audio content and subtitle content of the video are identified to obtain the video emotion coefficient; according to the video posting interaction data, the number of video interactions, the video interaction vocabulary and the video interaction emotion coefficient of the posting user are obtained to obtain the first interaction data; according to the video emotion coefficient, the posting video length and the first interaction data, the first posting coefficient is obtained; the video interaction emotion coefficient is obtained by identifying the interaction data of the posting user, including interactive comments, liked data and shared data, etc.

[0127] According to the interactive data of picture and text, the image content and text content of the picture and text are identified to obtain the picture and text sentiment coefficient; according to the picture and text interaction times, picture and text interaction vocabulary and picture and text interaction sentiment coefficient of the posting user in the picture and text posting interactive data, the second interactive data is obtained; according to the picture and text sentiment coefficient, the number of posted pictures and texts and the second interactive data, the second posting coefficient is obtained. The picture and text interaction sentiment coefficient is obtained by identifying the interactive data of the posting user, including interactive comments, liked data and shared data, etc.

[0128] S40. Identify the user location and the first cultural interest coefficient of the first user according to a clustering algorithm to obtain first distribution data of the first user in a geographical dimension.

[0129] The process of identifying and dividing user locations and first cultural interest coefficients through clustering algorithms is as follows:

[0130] A clustering algorithm based on density dimension is selected to identify user locations and first cultural interest coefficients, and the first cultural interest coefficient is used as the weight of the user location to act on distance correction and density identification between user locations;

[0131] User locations whose user location density is greater than a preset density threshold are classified into one category to obtain first distribution data.

[0132] The present invention identifies user locations and cultural interest coefficients based on the measurement dimension of user density through a clustering algorithm, uses the cultural interest coefficient as the weight of the user location, and acts on the distance correction and density identification between user locations; then divides the user locations whose user location density is greater than a preset density threshold into one category to obtain distribution data; accurately identifies and divides the user's density distribution to obtain distribution data of users interested in traditional culture.

[0133] S50. Obtain user data of the international network social platform in the second period to construct a second data set; identify the second data set according to the method of retrieving and identifying the first user, the first cultural interest coefficient and the first distribution data of the first data set to obtain the second user, the second cultural interest coefficient and the second distribution data.

[0134] S60. Obtain regional distribution data based on the first distribution data and the second distribution data; the regional distribution data includes a first distribution area, a second distribution area and a third distribution area; and obtain a cultural communication effect coefficient based on the user age, cultural interest coefficient, user density and regional area in the regional distribution data.

[0135] The process of obtaining the regional distribution data is as follows:

[0136] Overlap the geographical dimensions of the first distribution data and the second distribution data accordingly; take the overlapping area of ​​the first distribution data and the second distribution data as the first distribution area; take the area of ​​the first distribution data that does not overlap with the second distribution data as the second distribution area; take the area of ​​the second distribution data that does not overlap with the first distribution data as the third distribution area.

[0137] like Figure 3 As shown, the first distribution data 10 and the second distribution data 20 are obtained by identification; the regional distribution data 30 is obtained according to the first distribution data 10 and the second distribution data 20; the overlap of the first distribution data and the second distribution data in the regional distribution data 30 is identified to obtain the first distribution area 31, the second distribution area 32 and the third distribution area 33.

[0138] The present invention obtains first distribution data of a first period and second distribution data of a second period; overlaps the geographical dimensions of the first distribution data and the second distribution data accordingly; obtains regional distribution data based on the overlapping distribution of the first distribution data and the second distribution data; and accurately identifies the distribution changes of users between the first period and the second period.

[0139] According to the average age change between the first user and the second user, the cultural interest coefficient change, the user density change and the area of ​​the first region within the first distribution area, a first propagation coefficient is obtained;

[0140] The second propagation coefficient is calculated based on the average age of the first user, the first cultural interest coefficient, the user density, and the area of ​​the second region within the second distribution area;

[0141] The third propagation coefficient is calculated based on the average age of the second users, the second cultural interest coefficient, the user density and the area of ​​the third region within the third distribution area;

[0142] The formula of the cultural communication effect coefficient is:

[0143] CulF=α1*Eff1+α2*Eff2+α3*Eff3;

[0144] Among them, CulF represents the cultural communication effect coefficient; α1 represents the first communication weight; Eff1 represents the first communication coefficient; α2 represents the second communication weight; Eff2 represents the second communication coefficient; α3 represents the third communication weight; Eff3 represents the third communication coefficient;

[0145]

[0146] Among them, area1 represents the area of ​​the first distribution area; β1 represents the first age weight; represents the average age of the first users in the first distribution area in the first period; represents the average age of the second users in the first distribution area in the second period; AD represents the time difference between the second period and the first period; AT represents the age change threshold; γ1 represents the first cultural interest weight; represents the average cultural interest coefficient of the second users in the first distribution area in the second period; represents the average value of the first user's cultural interest coefficient in the first distribution area in the first period; CT represents the cultural interest coefficient threshold; η1 represents the first density weight; represents a second user density of the first distribution area in a second period; represents the first user density of the first distribution area in the first period; DT represents the user density threshold;

[0147]

[0148] Among them, area2 represents the area of ​​the second distribution area; β2 represents the second age weight; represents the average age of the first users in the second distribution area in the first period; γ2 represents the second cultural interest weight; represents the average value of the first user's cultural interest coefficient in the second distribution area in the first period; η2 represents the second density weight; represents a first user density of a second distribution area in a first period;

[0149]

[0150] Among them, area3 represents the area of ​​the third distribution area; β3 represents the third age weight; represents the average age of the second users in the third distribution area in the second period; γ3 represents the third cultural interest weight; represents the average cultural interest coefficient of the second user in the third distribution area in the second period; η3 represents the third density weight; It represents the second user density of the third distribution area in the second period.

[0151] The present invention obtains the user age, cultural interest coefficient, user density and area of ​​the first distribution area, the second distribution area and the third distribution area in the first period and the second period; calculates the cultural communication effect coefficient according to the changes in the user age, cultural interest and density from the first period to the second period; and accurately measures the cultural communication effect of the second period relative to the first period.

[0152] The present invention obtains user data of an international network social platform in a first period, constructs a first data set; retrieves first cultural social data and a first user according to a search index; constructs a cultural interest recognition model to recognize the first cultural social data, obtains a first cultural interest coefficient; recognizes the user position and the first cultural interest coefficient according to a clustering algorithm, obtains first distribution data; obtains user data of an international network social platform in a second period, constructs a second data set, recognizes a second user, a second cultural interest coefficient, and a second distribution data; obtains regional distribution data according to the first distribution data and the second distribution data, and calculates a cultural communication effect coefficient. The present invention accurately measures the communication effect of traditional culture through the cultural communication effect coefficient.

[0153] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for evaluating the effect of international communication of digital traditional culture based on deep learning, characterized in that: include: S10. Obtain user data of the international network social platform in the first period to construct a first data set; the user data includes user age, user location and user social data; the user social data includes browsing comment data and posting interaction data; S20. Determine a search index according to the type of traditional culture, and search the user social data in the first data set according to the search index to obtain the first cultural social data and the corresponding first user; S30. Constructing a cultural interest recognition model to recognize the first cultural social data, and obtaining a first cultural interest coefficient according to the first browsing comment data and the first posting interaction data in the first cultural social data; S40. Identify the user location and the first cultural interest coefficient of the first user according to the clustering algorithm to obtain first distribution data of the first user in the geographical dimension; S50. Obtain user data of the international network social platform in the second period to construct a second data set; identify the second data set according to the method of retrieving and identifying the first user, the first cultural interest coefficient and the first distribution data of the first data set to obtain the second user, the second cultural interest coefficient and the second distribution data; S60. Obtain regional distribution data based on the first distribution data and the second distribution data; the regional distribution data includes a first distribution area, a second distribution area and a third distribution area; and obtain a cultural communication effect coefficient based on the user age, cultural interest coefficient, user density and regional area in the regional distribution data.

2. According to the method for evaluating the effect of international communication of digital traditional culture based on deep learning in claim 1, it is characterized by: The cultural interest recognition model is constructed based on deep learning, and includes a browsing behavior recognition layer, a posting behavior recognition layer, and a cultural interest recognition layer; The browsing behavior recognition layer divides the browsing comment data according to the browsing content to obtain video browsing comment data and picture and text browsing comment data; A first browsing coefficient is obtained by identifying the video browsing comment data; a second browsing coefficient is obtained by identifying the picture and text browsing comment data; and a cultural browsing coefficient is obtained by calculating the first browsing coefficient and the second browsing coefficient; The posting behavior recognition layer divides the posting interaction data according to the posting content to obtain video posting interaction data and picture and text posting interaction data; and obtains a first posting coefficient according to the video posting interaction data; The second posting coefficient is obtained by identifying the interactive data of picture and text posting; the cultural posting coefficient is obtained by calculating the first posting coefficient and the second posting coefficient; The cultural interest identification layer calculates the cultural interest coefficient based on the cultural browsing coefficient and the cultural posting coefficient.

3. The method for evaluating the effect of international communication of digital traditional culture based on deep learning according to claim 2 is characterized by: The browsing behavior recognition layer recognizes video browsing comment data and picture and text browsing comment data in the following process: According to the video browsing comment data, the video browsing time of the user and the total time of the video, the first video data is obtained; according to the video browsing comment data, the number of video comments, the video comment vocabulary and the video comment sentiment coefficient of the user, the second video data is obtained; according to the first video data and the second video data, the first browsing coefficient is calculated; The first picture and text data is obtained based on the user's picture and text browsing time and the platform average browsing time of the pictures and texts in the picture and text browsing and commenting data; the second picture and text data is obtained based on the number of picture and text comments, the picture and text comment vocabulary and the picture and text comment sentiment coefficient in the picture and text browsing and commenting data; the second browsing coefficient is calculated based on the first picture and text data and the second picture and text data.

4. The method for evaluating the effect of international communication of digital traditional culture based on deep learning according to claim 2 is characterized by: The posting behavior recognition layer recognizes video posting interaction data and picture and text posting interaction data in the following process: According to the video posting interaction data, the image content, audio content and subtitle content of the video are identified to obtain the video emotion coefficient; according to the video posting interaction data, the video interaction times, video interaction vocabulary and video interaction emotion coefficient of the posting user are obtained to obtain the first interaction data; according to the video emotion coefficient, the posting video duration and the first interaction data, the first posting coefficient is obtained; According to the interactive data of picture and text posts, the image content and text content of the picture and text are identified to obtain the picture and text sentiment coefficient; The second interaction data is obtained according to the number of picture-text interactions, the picture-text interaction vocabulary and the picture-text interaction sentiment coefficient of the posting user in the picture-text posting interaction data; the second posting coefficient is obtained according to the picture-text sentiment coefficient, the number of posted pictures and texts and the second interaction data.

5. According to the method for evaluating the effect of international communication of digital traditional culture based on deep learning in claim 1, it is characterized by: The process of identifying and dividing user locations and first cultural interest coefficients through clustering algorithms is as follows: A clustering algorithm based on density dimension is selected to identify user locations and first cultural interest coefficients, and the first cultural interest coefficient is used as the weight of the user location to act on distance correction and density identification between user locations; User locations whose user location density is greater than a preset density threshold are divided to obtain first distribution data.

6. The method for evaluating the effect of international communication of digital traditional culture based on deep learning according to claim 1 is characterized by: The process of obtaining the regional distribution data is as follows: Overlap the geographical dimensions of the first distribution data and the second distribution data accordingly; take the overlapping area of ​​the first distribution data and the second distribution data as the first distribution area; take the area of ​​the first distribution data that does not overlap with the second distribution data as the second distribution area; take the area of ​​the second distribution data that does not overlap with the first distribution data as the third distribution area.

7. The method for evaluating the effect of international communication of digital traditional culture based on deep learning according to claim 1 is characterized by: According to the average age change between the first user and the second user, the cultural interest coefficient change, the user density change and the area of ​​the first region within the first distribution area, a first propagation coefficient is obtained; The second propagation coefficient is calculated based on the average age of the first user, the first cultural interest coefficient, the user density, and the area of ​​the second region within the second distribution area; The third propagation coefficient is calculated based on the average age of the second users, the second cultural interest coefficient, the user density and the area of ​​the third region within the third distribution area; The cultural communication effect coefficient is calculated based on the first communication coefficient, the second communication coefficient and the third communication coefficient.

8. According to claim 1, the method for evaluating the effect of international communication of digital traditional culture based on deep learning is characterized by: The formula of the cultural communication effect coefficient is: CulF=α1*Eff1+α2*Eff2+α3*Eff3; Among them, CulF represents the cultural communication effect coefficient; α1 represents the first communication weight; Eff1 represents the first communication coefficient; α2 represents the second communication weight; Eff2 represents the second communication coefficient; α3 represents the third communication weight; Eff3 represents the third communication coefficient; Among them, area1 represents the area of ​​the first distribution area; β1 represents the first age weight; represents the average age of the first users in the first distribution area in the first period; represents the average age of the second users in the first distribution area in the second period; AD represents the time difference between the second period and the first period; AT represents the age change threshold; γ1 represents the first cultural interest weight; represents the average cultural interest coefficient of the second users in the first distribution area in the second period; represents the average value of the first user's cultural interest coefficient in the first distribution area in the first period; CT represents the cultural interest coefficient threshold; η1 represents the first density weight; represents a second user density of the first distribution area in a second period; represents the first user density of the first distribution area in the first period; DT represents the user density threshold; Among them, area2 represents the area of ​​the second distribution area; β2 represents the second age weight; represents the average age of the first users in the second distribution area in the first period; γ2 represents the second cultural interest weight; represents the average value of the first user's cultural interest coefficient in the second distribution area in the first period; η2 represents the second density weight; represents a first user density of a second distribution area in a first period; Among them, area3 represents the area of ​​the third distribution area; β3 represents the third age weight; represents the average age of the second users in the third distribution area in the second period; γ3 represents the third cultural interest weight; represents the average cultural interest coefficient of the second user in the third distribution area in the second period; η3 represents the third density weight; It represents the second user density of the third distribution area in the second period.

Images