Using data to interpret "Ode to Joy 2" from all angles

As a data analysis enthusiast, the author of this article will interpret all aspects of the hit drama " Ode to Joy 2 " from a data perspective, including public opinion dissemination, online reputation, character social network analysis, and text analysis of the novel content.

On May 11, the second season of "Ode to Joy", which many Andy fans have been looking forward to for a long time, was finally aired. Compared with the first season, "Ode to Joy 2" appears more refined and fashionable in styling and scenes, and the styling of each character is in line with his or her identity and background, and is not abrupt.

As the saying goes, "the layman sees the excitement, the expert sees the doorway." As a data analysis enthusiast, the author naturally wants to interpret every aspect of this hit drama from a data perspective, including public opinion dissemination, online reputation, character social network analysis, and text analysis of the novel's content.

The following is what I want to focus on in this article:

• How does this part of public opinion spread on the Internet? Including the drama's overall network dissemination trends, portraits of the audience, and online marketing techniques.
• Compared with the hugely successful first part, how is the reputation of the second part? It mainly involves the text analysis of Douban film reviews of this drama.
• Analysis of the novel texts of the first and second seasons. It mainly involves keyword extraction, social network relationships of main characters and related word analysis, etc.

In the process of analysis, the author will use some commonly used (big) data analysis tools, as well as interlocking data analysis discussions. Although they are not directly helpful for conventional data operations, "a stone from another mountain can be used to polish a jade", and the analysis ideas can be used as a reference in order to be applied to other actual scenarios.

1 Big Data Public Opinion Analysis of Ode to Joy 2

In the first part, the author selected the online public opinion data about "Ode to Joy 2" from 2017.05.07 to 2017.05.29 as the data analysis object, and used Sina Micro Public Opinion's "Information Monitoring" and "Online Event Analysis" as tools for data collection, data analysis and visualization.

The following is an analysis of the overall network dissemination overview and dissemination trends of "Ode to Joy 2".

1.1 Overview of the Internet Distribution of Ode to Joy 2

(1) Trends in online communication

The above picture is the information dissemination trend chart of "Ode to Joy 2" during this period. Since the first day of broadcasting, the drama's overall online voice has increased significantly, and there have been several "peaks" of information dissemination, indicating that there were more discussions about the drama on the entire network that day (including media and individual netizens). It is particularly noteworthy that the peak value on the day the drama was first broadcast (5-11) was the largest in that time period. Is this a coincidence? Don’t worry, I will reveal the answer to you later.

(2) Proportion of positive and negative information on the entire network

As can be seen from the above figure, during this period, the total amount of information about "Ode to Joy 2" on the entire network was 5,564,016, of which 152,315 were negative information, accounting for 2.74% of the total amount of information.

At the same time, overall, negative information tends to increase over time.

Combining the two, such a proportion and development of negative information is not favorable for the drama, indicating that the negative word-of-mouth evaluation of the drama on the entire network is quite significant, which needs to attract the attention of the producers. In the following analysis, the author will find out what negative word-of-mouth has appeared and the reasons for their occurrence.

(3) Sources of attention across the entire network

Because Weibo accounts for a large proportion of the dissemination of film and television dramas, the author excluded Weibo as an information source and then conducted a statistical analysis of the remaining TOP10 information sources, as shown in the following figure:

As can be seen from the above figure, in addition to Weibo, WeChat and Toutiao, the two major social media platforms, contribute the most information about the drama. In particular, WeChat recently launched the "WeChat Lab", which allows users to actively search for keywords through the newly added "Search" and "Take a look" entrances, and browse hot information, information that friends are paying attention to, and articles that they are interested in. Such actions, coupled with the accumulation of high-quality content over the years, have made WeChat's importance in social media content production increasingly prominent.

(4) Geographical distribution of network information

It's a bit surprising that Jiangsu ranks first in the list of online attention (the amount of information about the drama on the entire network), and it's not Beijing and Shanghai, two modern cities with a large number of young white-collar workers. Shanghai in particular ranks last in the TOP10 in terms of the amount of information about the drama. Could it be that most of the netizens and media in Shanghai have given up on the drama? Or is it that the show has a very bad reputation?

The author will analyze this question in the second part.

1.2 Crowd Profile

Because Weibo (mainly Sina Weibo) accounts for the majority of discussions about this drama, the author chose people on this platform as the subject of analysis.

(1) Gender distribution

There is no doubt that girls are loyal fans of the show and their enthusiasm for it is definitely greater than that of boys.

(2) Age distribution

As can be seen from the figure below, people in the age range of 19-34 account for about 70% of the audience, which means that young people are very interested in the drama.

(3) Interest Graph

The figure below shows the TOP5 interest tags of people who follow the drama. It can be seen that this group of people are more star-chasing and love life, which are interests and hobbies that are very characteristic of young people.

(4) User emotional state

Based on the emotional category labels given to themselves by Weibo users who follow Ode to Joy 2, we can learn the emotional state distribution of these users:

From this we can see that most users (80.75%) are in an emotional blank/open window period ("single" and "seeking relationship" status), followed by "in love". Combined with the previous age distribution, it shows that the group of people paying attention are mostly unmarried young people.

(5) Educational background distribution of users

There is no doubt that the audience of this type of urban emotional drama is mainly people with higher education, because the drama tells the story of highly educated white-collar workers working, living and falling in love in big cities.

(6) User city type

It is a bit surprising that people from 4th tier and below cities account for the highest proportion of 30.26%, followed by 2nd and 3rd tier cities, while people from 1st tier cities account for only 14.41%. Could it be that people in 1st tier cities are more picky about this drama?

Combined with the regional distribution of the previous information, the main audience of "Ode to Joy 2" is a group of highly educated young women who love life and are keen on chasing stars. They live outside the first-tier cities and are mostly single.

1.3 Analysis of the marketing promotion of Ode to Joy 2 on its first broadcast day

In the aforementioned network-wide communication trend, the author mentioned that the drama had an obvious peak in volume on the day it was aired, and it is obvious that this is not accidental.

From the distribution of information sources above, we can see that most of the voice of the drama comes from Weibo, especially Sina Weibo. For this purpose, the author separately retrieved the Weibo information trend data from 5-11.

As can be seen from the above figure, the amount of information about the drama on Sina Weibo is generally on an upward trend. It reached its peak around 23:00 that day, and the growth trend was very rhythmic - starting from 9:00 during working hours, the volume continued to increase over time.

In order to further discover the dissemination patterns, the author used the "Information Monitoring" module of Sina Micro Public Opinion to further search and investigate the influential Weibo posts, that is, to find the Weibo posts with more forwarding times, and to trace the source of the dissemination.

After gradually unraveling the mystery, the author found several Weibo posts with the largest number of reposts, which are the sources of dissemination. They are mainly the Weibo accounts of several leading actors of "Ode to Joy 2" and some Weibo topics related to the drama. Judging from the actual results, several leading actors of "Ode to Joy 2" - Liu Tao, Wang Kai, Yang Zi and others have a high influence on Sina Weibo, with a large number of fans and active forwarding, which makes the effect of promoting the series relatively good.

In addition to the active promotion by the main creative staff, "Ode to Joy 2" also has rich and diverse promotional and marketing methods. In terms of promotion on Weibo, Ode to Joy 2 had sufficient promotional materials, including trailers, tidbits, theme song MVs, etc.; the promotional matrix was complete, not only paying attention to the linkage with official Weibo accounts such as "Sina Entertainment", "Sina TV", "Dragon TV Tomato Channel" and "@Weibo TV Series", etc.), advertising brands, but also investing heavily in KOL public opinion guidance, and inviting other celebrities to help with promotion (such as Fan Bingbing).

However, judging from the communication effects above, the communication effect of large media accounts is not as great as that of celebrities’ personal Weibo accounts. So, what is the reason? Let’s continue reading.

1.4 Analysis of the Communication Effect of Weibo

In order to gain a deeper understanding of the characteristics and patterns of communication on Weibo by major media accounts and celebrities, the author specifically found two Weibo posts with the largest amount of fan interaction on May 11 - @微博电视 and Liu Tao's personal Weibo account @刘涛tamia. Their reposted comments were almost the same, making it easy to discover the communication patterns under the same conditions.

The forwarding and propagation relationship data of these two Weibo posts are exported to create a visual “force-directed propagation diagram” that can clearly reflect the propagation path:

The picture above is a communication path diagram of @微博电视's marketing Weibo about "Ode to Joy 2". The center of the purple "cluster" is the account @微博电视. In addition to the forwarding communication brought by it (that is, the small dots in the purple cluster area), there are some secondary communication centers around it, such as "十三点半的kings", "影影大米FAN", "吴坎坎kkw" and other self-media accounts with strong marketing nature to assist in promotion. Obviously, this is traffic bought by the promoter, not self-generated.

The following Weibo post by Liu Tao shows an obvious " centralized " feature, that is, apart from Liu Tao's personal Weibo account, there is no particularly obvious secondary communication center, and the vast majority of the communication comes from Liu Tao's account fans.

From this, the conclusion is self-evident:

The dissemination of celebrities' personal Weibo is more effective than that of large media accounts, because celebrities attract many fans due to their personal charm, the Weibo posts are close to their lives, and they have a strong interactive relationship with their fans. In addition, their fan groups will spare no effort to spread such news, so the dissemination effect is stronger. In contrast, due to the marketing attributes of large Weibo accounts, fans will not actively participate in the interaction unless there is a profit motive. Generally speaking, they will not be too strong. What's worse, they will mobilize water armies to operate and seek benefits through improper means.

Well, the above is the big data public opinion communication analysis of "Ode to Joy 2". Next, we will enter another section - text mining of the drama's reputation (mainly Douban film reviews) and the novel text, and look forward to discovering some more interesting content.

2 Analysis of Douban’s reputation and novel text of Ode to Joy 2

Generally speaking, in-depth exploration of texts can often achieve the effect of "knowing what it is and why it is", because texts are not as empty and abstract as numbers, and the content can often directly reveal the reasons.

The data in this part mainly include:

• Douban movie reviews of Ode to Joy Season 1 and Season 2
• The original novels of the first and second seasons of "Ode to Joy" will be packaged and filmed in "Ode to Joy 2", so the content of the second and third seasons of the novels will appear in the plot of "Ode to Joy 2".

Douban Movies is China's largest and most authoritative film sharing and review community, with millions of pieces of information on films and filmmakers, more than 2,500 cinemas and tens of millions of people who love movies. Therefore, the reviews of movies and TV series here are representative and credible, and can largely reflect the reputation of a film or TV series. Therefore, the author captured the reviews of Ode to Joy Season 1 and 2 on Douban as the basis for word-of-mouth analysis.

2.1 Comparative analysis of the overall reputation of Ode to Joy Season 1 and Season 2

(1) Rating star

Judging from the overall score, the second season of "Ode to Joy" is obviously inferior to the first season. The former received a high score of 7.3, while the latter only received a score of 5.2.

From the star rating ratio of the first and second seasons, we can see that the reason why the second season has a lower overall score than the first season is that the second season accounts for a smaller proportion of the mid-to-high score range of "4 stars" and "3 stars" than the first season, while the proportion of the low score range of "2 stars" and "1 star" is higher than the first season.

(2) Analysis of the sentiment tendency of Douban film reviews of Ode to Joy 2

Since the first season has a better reputation than the second season, the author will focus on analyzing the "emotional tendency" of Douban movie reviews for the second season.

The "emotional tendency" here is represented by an "emotional polarity value" between 0 and 1. If the emotional polarity value of a comment is closer to 0, it means that its negative emotional tendency is more obvious, and the closer it is to 1, the more obvious its positive emotional tendency is. If the value hovers around 0.5, it means that the emotional tendency is not obvious and is "neutral". The author subdivides the emotional tendency value interval of 0~1 into three sub-intervals, corresponding to different emotional tendencies:

Douban movie reviews have a unique language style - irony, using both negative and positive terms. Unlike product reviews that are straightforward and explicit, their semantics are often implicit and humorous. Therefore, before the analysis, the author found a large amount of annotated Douban movie review corpus and used machine learning to train the sentiment semantic model in order to achieve a good effect of positive and negative sentiment judgment. The preliminary results are as follows:

After processing the sentiment label mapping, the following results are obtained:

As can be seen from the above figure, the biggest difference in the emotional tendencies of the first and second seasons lies in the different proportions of positive and negative emotions: the proportions of positive and negative emotions in the comments of the first season are evenly matched, while in the second season, negative emotions clearly dominate, and the proportion of positive emotions is greatly compressed, which is one of the reasons why the Douban score of this drama is not high.

Let’s take a look at the trend of emotional tendencies over time. The analysis period selected by the author is March 31st to May 22nd, which is the trend of emotional tendencies more than a month before the show was broadcast and a period of time after the show was broadcast. The value used here is the cumulative value of sentiment tendency. The higher the value, the better the word-of-mouth evaluation. This also takes into account the increase or decrease in the number of comments (the willingness of Douyou to participate in the rating, even if it is a negative rating, shows that there is a certain degree of attention to the show). This gives us the following figure:

It can be seen that the emotional tendency value reached its peak on the day after the show was broadcast, and then declined all the way. Combined with the overall emotional tendency proportion during this period, it can be seen that the attention and word-of-mouth evaluation of the series after the broadcast of the show showed a downward trend.

So, what is the reason why the majority of fans dislike this drama so much? Please continue reading.

(3) Comparative analysis of keywords in Douban comments on Ode to Joy Season 1 and Season 2

The following is a comparative analysis of keywords in Douban’s comment texts on Ode to Joy Season 1 and Season 2. In this intuitive visualization chart, you can see Douban’s comments on all aspects of Ode to Joy Season 1 and Season 2.

In the above figure, the comment texts for seasons one and two are divided into "frequent" and "infrequent" respectively. The above picture can be divided into 4 quadrants based on the "Average" on the horizontal and vertical axes. The upper right corner contains words with high mention rates in both types of texts, and the lower left corner contains words with low mention rates in both types of texts. The top left and bottom right words have only one of the two high mention rates.

In addition, the words represented by the blue dots belong to the evaluation of the first season of "Ode to Joy", and the words represented by the yellow dots belong to the evaluation of the second season of "Ode to Joy". The comments of the two in the upper right quadrant tend to overlap.

In the TOP Terms column on the right, we can see that in their comments on the first season, the main things mentioned were the cast (strong cast), Shanying (Shandong Film and Television Production Co., Ltd., the show's producer), being unable to stop watching the show, the richness of the characters, and the discussion of the emotional thread between Andy and Singularity (Wei Wei). However, there were no obvious negative comments, and the comments were mostly positive.

In the second season, many netizens compared the first and second seasons, and thought that this drama was incomparable to the first season. Also, there were too many interludes in this drama, which was like a collection of MVs ("MV", "mv", "song", "just sing"). As one commentator said, "There were a lot of BGMs from the beginning, and it didn't seem to fit the plot. There was a country vibe to the curry song." . . It is interesting to note that some people have compared this drama with "White Deer Plain" which was produced in the same time period. Both dramas contain discussions on the era environment and individual development, but they are quite different in structure, vision and final presentation.

Click on the keywords to see the comment details about these keywords (the sentences are segmented and stop words are removed).

In short, fans have a lot to complain about the second season.

Therefore, the author would like to take a further look at what other complaints the netizens have about the drama.

(4) Comparative analysis of typical opinion mining in the second season

The 5,000 typical comments obtained after preprocessing are clustered. Each text clustering "cluster" will have a central sentence, which is the typical opinion of the "cluster". Therefore, this kind of text processing is also called "typical opinion mining", which is used to discover the main viewpoints or opinions from massive texts.

After processing by the author, 5,000 comments were divided into more than 300 "topic clusters" with different proportions. The author only selected the top 10 typical opinions for display.

As can be seen from the table above, the typical opinions of the soybean oil fans about this drama are mostly negative, focusing on the following aspects:

• Worse than the first part, the main criticisms are the plot, the screenwriter and the actors' acting skills, as well as the abrupt insertion of character-specific interludes/music
• The three values ​​are not right, and they are complained as "a high-end version of Tiny Times", advocating the values ​​of the rich who "live a life of luxury and extravagance".
• Preference for individual characters and their leading actors in the play, such as Qiu Yingying and Zhao Qiping and their actors

In addition, many people complained that the show had too many product placements. In order to accommodate the product placements, the food, clothing and daily expenses of the characters were obviously inconsistent. Some people even said that "Ode to Joy 1 inserted the right amount of advertisements into the plot, while Ode to Joy 2 inserted the plot into the advertisements..."

(5) Word cloud of comments on the “Five Beauties”

Based on the comments of netizens about the "five beauties" in the first and second seasons of "Ode to Joy", the author created a personalized comment keyword cloud for Andy, Qu Xiaoxiao, Fan Shengmei, Guan Juer and Qiu Yingying.

Without further explanation from the author, you can see that these character keyword clouds mainly reflect the readers' impressions of the "Five Beauties" and their main related characters. The personality traits of these five people can be summarized as follows from the above keyword word cloud:

• Andy, the iceberg queen with a strong aura and cold exterior
• The quirky and charming little fairy - Qu Xiaoxiao
• The vain, materialistic, independent and proud queen - Fan Shengmei
• Guan Juer, a quiet, restrained and harmonious girl
• Qiu Yingying, a simple, cute and energetic girl

2.2 Analysis of the novel text of Ode to Joy Season 1 and 2

(1) Comparison of keywords in the novel texts of Season 1 and Season 2

Here, for keyword extraction of novel text, the author did not adopt the word frequency statistics used in general text analysis, but adopted the TF-IDF (term frequency–inverse document frequency) keyword statistics method, so as to more effectively extract key information from the text. For details, please refer to the third part of "Taking "The Rise of the Great Qin Empire" as an example to discuss big data public opinion analysis and text mining."

The following are the top 20 keywords of the original novels of Ode to Joy Seasons 1 and 2.

It can be seen that the keywords of the first and second seasons of the novel are mostly the main characters, and the overall order of the main characters (the importance of the characters for the novel) has not changed much, and the importance of some individual characters has "shifted":

• Andy, Qu Xiaoxiao and Fan Shengmei are the first, second and third female protagonists in seasons one and two;
• Bao Yifan's importance increases in the second season, while Wei Wei's importance decreases;
• Xie Bin (also known as Xie Tong, a music expert in the TV series), Shu Zhan and Chen Jiakang, who appeared in the second season of the TV series, also appeared in the top 20 keywords of the second and third seasons of the novel, but their rankings were relatively low.

This also shows that the novel and the script maintain consistency in character settings, with no obvious major changes.

(2) Analysis of social network of characters in the first and second seasons

By setting up a character dictionary, the author extracted the character social relationship map of the novel text corresponding to the first and second seasons of "Ode to Joy" to discover the characteristics of the character relationships in the two TV series.

In a character social network, each node represents a character, and the lines represent the relationships between characters. Character nodes with different line colors represent that they have more frequent interactions (either friends or enemies). The font size of the nodes and their representative characters represents "Betweenness Centrality". The academic term for this term is "the interaction between two non-adjacent members depends on other members in the network, especially those members on the path between the two members, who have some control and constraint on the interaction between the two non-adjacent members." In layman's terms, characters with large fonts have greater influence on interpersonal relationships and can contact and influence more people. However, characters who appear frequently in the novel are not necessarily of this type. The sense of presence here needs to be based on connections and influence. For details, please refer to the third part of "Taking "The Rise of the Great Qin Empire" as an example to discuss big data public opinion analysis and text mining."

As can be seen from the above figure, Andy in the first season has strong interpersonal connections with several main characters, and the lines between them are thicker, while Tan Zongming, Qu Xiaoxiao, Fan Shengmei and others have rich network resources, or they have more interpersonal interactions/social activities.

In the social network of characters in the second season, there is a trend of "decentralization". Except for Bao Yifan who becomes a lover, the first protagonist Andy's interactions with other people are not as close as before. As the plot develops, new emotional possibilities emerge between them, but they also face new troubles brought about by their background and the Bao family. In addition to the constant blame from her stepmother, Mrs. Wei, it is not difficult to understand why Andy interacts less with others than in the first season, as he has constant troubles of his own and has no time to take care of them.

Of course, the emotional context of the characters in the second season is also very clear. Even with the emergence of troublemakers (Wei Wei, Shu Zhan, Chen Jiakang, etc.), the fixed CP combinations have not changed, such as Andy-Bao Yifan, Qiu Yingying-Ying Qin, Qu Xiaoxiao-Zhao Qiping, Fan Shengmei-Wang Baichuan, and Guan Juer-Xie Bin (Xie Tong).

(3) Analysis of Character-Related Words in the Novel Text

Finally, the author combined seasons 1, 2, and 3 of the novel Ode to Joy, and after text preprocessing, performed word vector training on the novel text, mapping these words into the vector space and turning them into word vectors (Word Vector) so that these words can be easily recognized and analyzed by computers while also having semantic relevance, rather than just being based on the co-occurrence relationship between words.

Due to space constraints, the author only analyzes the associated words related to the "Five Beauties", presenting the TOP10 associated words for each category, thereby discovering the vocabulary related to the main characters and exploring the deeper characteristics of the characters.

From the TOP10 related words of "五美" above, we can get two types of related words:

The first category of words refers to characters that are closely related to the main character , such as Wei Wei and Bao Yifan to Andy, Zhao Qiping and Liu Xinhua to Qu Xiaoxiao, etc. These are in line with everyone's understanding of the TV series.

The other type is words that reflect a character's behavior . For example, Andy and Qu Xiaoxiao both use telephones as their common means of communication, while the other three only use mobile phones. A telephone generally refers to a landline phone, which is located in an office and is more formal and is used more often by leaders or management. Mobile phones, on the other hand, are portable and are used more often both indoors and outside.

The associated words for Andy, Qu Xiaoxiao and Fan Shengmei include "last night". When comparing with the original novel, the three of them had many social events at night.

In addition, the word "连忙" appears in the conjunctions of Fan Shengmei, Guan Juer and Qiu Yingying, which means "immediately" or "right away", which means that they often appear in a hurry and in a hurry.

Every word reflects the social class and mentality of the "Five Beauties".

Conclusion

Although after analysis, it is found that "Ode to Joy 2" has many flaws in the plot and various details, coupled with the very incongruous hard advertising implantation, its reputation is far inferior to the first season. However, "Ode to Joy 2" still inherits the typical character settings of the first season relatively well. Many viewers can see the shadows of themselves or people around them through the "five beauties". Although it is a novel, it also closely observes the current reality and has a natural sense of familiarity and substitution. It can also trigger emotional resonance between the drama and the audience, and is even regarded by many urban white-collar workers as a survival guide and a textbook for love and mate selection.

Perhaps, like many TV dramas that have received mixed reviews, the judgment on them is inherently ambiguous and cannot be concluded immediately:

A film that is praised by the world may not be a bad film; a film that is criticized by the world may not be a bad film.

So, friends who have read this article or watched this drama, what do you think?

ps: When I finished writing this article, I hadn’t watched a single episode of the second season. If I have misinterpreted the plot, I hope senior drama fans will forgive me, hahahaha.

References and tools:

1. Sina Micro Public Opinion: "Network Event Analysis", "Information Monitoring" and "Microblog Communication Effect Analysis" modules (data collection, data analysis and data visualization)
2. Micro Index, Weibo Data Center Bingjian User Insight Platform (Character Portrait)
3. Original novels of Ode to Joy Seasons 1, 2, and 3 (text data)
4. Douban movie reviews of Ode to Joy 1 and Ode to Joy 2 (rating data and text data)
5. Python, jieba word segmentation, snownlp, gensim, tagul (text segmentation, sentiment analysis, word vector and word cloud production)

Mobile application product promotion service: APP promotion service Qinggua Media advertising

This article was compiled and published by @Scottish Fold Ear Cat (Qinggua Media). Please indicate the author information and source when reprinting! Site Map