Comprehensive Evaluation Model of Network Drama Hotness Based On

2019 5th International Conference on Economics and Management (ICEM 2019) ISBN: 978-1-60595-634-3

Comprehensive Evaluation Model of Network Drama Hotness Based on Factor Analysis Xin-Xing ZHAO1,a,* and Pei-Yi SONG1,b 1School of Economics and Management, Communication University of China, Beijing, China [email protected], [email protected] *Corresponding author

Keywords: Network Drama Hotness, Factor Analysis, Comprehensive Evaluation Model.

Abstract. Firstly, this paper constructed an index system for the comprehensive evaluation of network drama hotness from the two dimensions. They were characteristics of commercial success and characteristics of the popularity of network dramas. Then it used the factor analysis method to verify the rationality and validity of the index system from an empirical perspective, established the weight of two common factors, and obtained the evaluation model of comprehensive factor scores. The conclusion of empirical analysis showed that 62.373% of network drama hotness come from the factor of characteristics of commercial success and 29.287% come from the factor of characteristics of the popularity.

Introduction The main quality measurement index of traditional TV series is audience rating [1]. With the advent of the multi-screen era, view counts have gradually become an important indicator to measure the influence of TV series or network dramas. However, neither audience rating nor view counts is enough to objectively and comprehensively measure the hotness of TV series or network dramas. Many scholars have conducted researches on this issue. For example, Chunlian Liu interprets the reasons for the imbalance between word of mouth and audience rating of TV series the Rise of Phoenixes from producer, audience and communication environment [2]. Bing Hu and others carry out an empirical analysis of the influence of microblog on TV ratings due to the coincidence between TV audiences and microblog users [3]. Conglu GUI points out that once the “flow is greater than everything” becomes a prominent tendency, it will inevitably lead to the impetuous atmosphere of making click volumes or view counts the sole criterion, and even mislead some people to resort to fraud for profit [4]. Yong Zhou and others take traditional ratings, time-shifted ratings, Internet clicks, online public opinions into consideration, and establish an evaluation index system for the effect of audio-visual information transmission [5]. After the video website platform downplays view counts, the hotness measurements launched by various companies lack neither unified industry measurement standards nor independent third-party regulatory agencies. Shortly after the announcement of formal closure of front-end view counts by Iqiyi, Guduo media and Entgroup add the corresponding indexes. This paper compares the compositional dimensions of content hotness, whole network hotness, and screening index, which are respectively introduced by Iqiyi, Guduo media and Entgroup. In these various measurement indexes, there are advantages and disadvantages in each other. However, they all lack transparent norms and authority. So it is a long way to stop making view counts the sole criterion. Whether from the theoretical discussion or the hotness index proposed by the industry, we can see that the ratings or view counts are one-sided. Especially in the process of network drama prediction, we need to consider the data of scores, audiences and website indexes. Therefore, this paper proposes a new definition of hotness value, which is used to predict the ranking of network dramas and achieve the purpose of relative accuracy. Based on the comprehensive analysis, this paper establishes a comprehensive evaluation model of network drama hotness. According to the hotness of network drama and its related evaluation indicators, the paper makes a quantitative analysis through factor analysis method, then makes a comparative analysis with the traditional evaluation method, and finally obtains more scientific and reasonable evaluation results.

43 Establishment of Index System How to quantitatively evaluate the hotness of network dramas and establish an objective evaluation system? Currently, there is no standard index system in the academic circle. There should be sufficient basis for the establishment of the index system. As an independent third-party data service provider of film and television entertainment industry, Entgroup launches screening index, which mainly reflects the comprehensive evaluation of the content value after a film or TV content is showed. It is composed of media-hotness, user-hotness, praise-degree and view-degree, but it has not built a complete index system of the first, second and third levels. Based on this, the paper takes media-hotness, user-hotness, praise-degree and view-degree as secondary indicators, and further refines them into the tertiary index. It also defines characteristics of commercial success and characteristics of the popularity as the primary index of the comprehensive evaluation of network drama hotness. (1) Characteristics of commercial success. The following indicators were selected as the secondary indicators to reflect the characteristics of commercial success: ①view-degree, ② user-hotness, ③media-hotness. Then these secondary indicators were further processed, and the following indicators were selected as the third-level indicators: ①single-year view counts, ②peak of Baidu “search index”, ③peak of Baidu “information index”, ④peak of Baidu “media index”. The corresponding relationship between third-level indicators and second-level indicators was presented in Table 1. (2) Characteristics of the popularity. The praise-degree indicator was selected as the secondary indicators to reflect the characteristics of the popularity. Then the secondary indicator was further processed, and the following indicators were selected as the tertiary indicators: ①score on the video platform, ②score on the Douban. The results were presented in Table 1.

Selection of Research Objects and Evaluation Indicators This paper collected the top 50 network dramas on the annual view counts ranking list in 2018 as samples. In these 50 network dramas, the single-year view count of Nirvana in Fire Ⅱ was missing, KO ONE RE-CALL has been removed, scores on the video platform of Let's Shake it 2 and MR Swimmer were missing, so these four network dramas were eliminated. Finally, there were 46 network dramas that participated in the estimation. The data of single-year view counts came from website Guduo media; peak of Baidu “search index”, peak of Baidu “information index” and peak of Baidu “media index” came from website Baidu index; score on the video platform came from website Iqiyi, Tencent, Youku, and so on; score on the Douban came from website Douban. The 46 network dramas were extracted according to the requirements of the comprehensive evaluation index system of network drama hotness. Thus, a 46*6 original data matrix, namely Xij, was obtained, in which i represented a network drama and j represented a third-level indicator.

Model Establishment Based on Factor Analysis of Hotness Index This paper used factor analysis to quantitatively observe the network drama hotness. Here we used SPSS to solve it. The process was as follows: (1) Suitability test. In this paper, the collected data was tested by KMO and Bartlett’s test. The results were presented in Table 2. In table 2, KMO=0.789>0.7, which indicated that factor analysis was suitable. The Sig value of Bartlett’s test was 0.000<0.05, so the null hypothesis was rejected. It showed that the variables were not independent, and there was correlation between variables. This ensured that the above indicators were suitable for factor analysis model.

44 Table 1. Comprehensive Evaluation Index System of Network Drama Hotness. First-level Second-level Third-level Description indicators indicators indicators Single-year Annual view counts of each network drama in View-degree view counts 2018 [one hundred million] Peak of Baidu Expressing netizens’ active search for a “search index” network drama in Baidu [ten thousand]

Data obtained by processing the number of Characteristics of User-hotness Peak of Baidu behaviors of netizens' reading, commenting, commercial “information forwarding, and likes. Measuring netizens’ success index” passive attention to a network drama [ten thousand] Peak of Baidu Number of news related to a network drama Media-hotness “media index” reported by major Internet media Score on the Score of a network drama on the video website Characteristics of video platform Praise-degree the popularity Score on the Score of a network drama on the Douban Douban

Table 2. KMO and Bartlett's Test. Kaiser-Meyer-Olkin Measure of Sampling Adequacy. .789

Bartlett's Test of Sphericity Approx. Chi-Square 320.426 df 15 Sig. .000 (2) Communality test. Communalities of variables reflected the degree to which all common factors explained the variance (variation) of original variables. The results were shown in Table 3. Communalities of all the variables were higher than 0.8, indicating that extracted common factors basically reflected more than 80% of original variables. The effect of factor analysis was good. (3) Selection and interpretation of common factors. In this paper, the total variance table was calculated by principal component analysis. The results were presented in Table 4. Two common factors were extracted, which coincided with the two dimensions of the index system proposed in this paper. Further analysis showed that the two common factors could explain 62.373% and 29.287% of hotness-related information respectively. Finally, the cumulative amount explained 91.659% of the overall information. The results showed that the two common factors could reflect well the overall information of network drama hotness. Because the explanation of initial factor load was not clear enough, it used variance maximum method to calculate the factor load matrix after rotation. The results were presented in Table 5.

Table 3. Communalities.

Initial Extraction Single-year view counts 1.000 .938 Peak of Baidu “search index” 1.000 .911 Peak of Baidu “information index” 1.000 .952 Peak of Baidu “media index” 1.000 .969 Score on the video platform 1.000 .866 Score on the Douban 1.000 .863

45 Table 4. Total Variance Explained. Component Initial Eigenvalues Rotation Sums of Squared Loadings Total % of Variance Cumulative % Total % of Variance Cumulative % 1 3.831 63.851 63.851 3.742 62.373 62.373 2 1.668 27.808 91.659 1.757 29.287 91.659 3 .274 4.570 96.229

dimension0 4 .124 2.070 98.299 5 .068 1.140 99.439 6 .034 .561 100.000

Table 5. Rotated Component Matrix. Component

1 2 Single-year view counts .956 .154 Peak of Baidu “search index” .952 .076 Peak of Baidu “information index” .970 .101 Peak of Baidu “media index” .984 -.023 Score on the video platform .044 .929 Score on the Douban .101 .924 From Table 5, the following information could be summarized: the first common factor had high correlation with single-year view counts, peak of Baidu “search index”, peak of Baidu “information index” and peak of Baidu “media index”; the second common factor had strong correlation with score on the video platform and score on the Douban. Comparing with the index system constructed before, we found that results of factor analysis coincided with the hotness index system. It took single-year view counts, peak of Baidu “search index”, peak of Baidu “information index” and peak of Baidu “media index” as a common factor F1. F1 represented characteristics of commercial success of network drama and explained 62.373% of the information. It took score on the video platform and score on the Douban as a common factor F2. F2 represented characteristics of the popularity and explained 29.287% of the information. (4) Comprehensive evaluation model. According to the component score coefficient matrix (see Table 6), we got score functions of each common factor, and then calculated scores of each common factor for network dramas.

Table 6. Component Score Coefficient Matrix. Component

1 2 Single-year view counts .253 .026 Peak of Baidu “search index” .257 -.020 Peak of Baidu “information index” .260 -.006 Peak of Baidu “media index” .272 -.080 Score on the video platform -.050 .541 Score on the Douban -.034 .534 F1=0.253X1+0.257X2+0.260X3+0.272X4-0.050X5-0.034X6 F2=0.026X1-0.020X2-0.006X3-0.080X4+0.541X5+0.534X6 Then we weighted and summarized the influence of variance contribution rate of each common factor on network drama hotness. Finally, we obtained the model of comprehensive factor F: F=(0.62373*F1+0.29287*F2) ∕0.91659 According to the above evaluation model, we got the top 10 of network dramas in 2018. The results were presented in Table 7. We could not only get the score and ranking of the comprehensive factor of each network drama, but also observe the score and ranking on each

46 common factor. For example, scores of the comprehensive factor F of Story of Yanxi Palace and Empresses in the Palace respectively reached 3.16 and 3.02, while comprehensive scores of other network dramas were not very different. We observed the ranking of the two common factors F1 and F2, and found that Empresses in the Palace was the only network drama which obtained both commercial success and the popularity. In addition, from the perspective of comprehensive ranking in the new order, the new ranking of network drama hotness has made some modifications to the original ranking according to view counts. For example, Growling Tiger, Roaring Dragon and The Words of Love had shot up the new ranking significantly compared with the original ranking. From the changes of rankings in Table 7, we could see that the index system and model adopted in this paper were more conducive to the ranking improvement of network dramas which had outstanding characteristics of the popularity.

Conclusion Firstly, this paper constructed an index evaluation system for network drama hotness based on the theoretical analysis, then collected the data of 50 network dramas in 2018 according to the index system, and finally conducted an empirical study on the index system with factor analysis method. From the results of analysis, we found that the factor analysis method had its unique advantage. First of all, Story of Yanxi Palace and Empresses in the Palace were much higher than other network dramas in terms of comprehensive scores, and they were the best network dramas in 2018. Secondly, according to their scores and rankings in each common factor, Story of Yanxi Palace was slightly better than Empresses in the Palace in terms of characteristics of commercial success, while Empresses in the Palace gained better reputation than Story of Yanxi Palace in terms of characteristics of the popularity. Finally, we found that network dramas represented by Growling Tiger, Roaring Dragon and The Words of Love were not equal in commercial success and the popularity, but they squeezed into the top 10 in the newly comprehensive ranking. Among the six excellent network audio-visual works released by the State Administration of Radio and Television in 2018, Growling Tiger, Roaring Dragon and The Words of Love were both selected as the outstanding network dramas of the year. These two dramas were not only excellent network dramas determined by the expert review committee, but also widely loved by the audience. Their scores on the Douban were all above 8 points. It can be seen that the final comprehensive evaluation results of this paper are consistent with the official selection and market feedback. It shows the rationality of comprehensive evaluation model. Therefore, the comprehensive evaluation model of network drama hotness can find high-quality network drama more objectively, comprehensively and scientifically. There is still room for further improvement in this paper. Firstly, the amount of data used in this model is relatively limited, and more data of network dramas are needed to verify and revise the model in the future. Secondly, this paper only examines the hotness of network dramas, and we need to conduct further comparative research on the source composition of the hotness of network dramas with different themes.

Acknowledgment Supported by the High-tech Project of Communication University of China (CUC18A015-2), Research on the Copyright Value Evaluation System and Model of TV Dramas and Network Dramas Based on Big Data Analysis. I wish to thank the anonymous reviewers and editors for their encouraging and insightful comments and suggestions. I am particularly grateful to Pei-Yi SONG, who encouraged me to write this paper, and gave me valuable comments and insightful feedback on earlier versions.

47 Table 7. Top 10 of Comprehensive Scores. Ranked Comprehensive Ranking Ranking Name F F F by view ranking 1 of F 2 of F 1 2 counts Story of 1 Yanxi 3.16 4.63 1 0.04 25 1 Palace Empresses 2 in the 3.02 4.01 2 0.90 8 2 Palace Agni 3 0.67 1.13 3 -0.31 32 3 Cantabile Martial 4 0.40 0.54 4 0.10 23 6 UniverseⅠ 5 Sand Sea 0.40 0.44 6 0.32 15 5 Eagles And 6 0.38 -0.35 26 1.92 1 8 Youngster Growling Tiger, 7 0.34 -0.30 22 1.70 3 17 Roaring Dragon The Legend 8 0.32 0.40 7 0.15 22 4 of Dugu 9 Ever Night 0.22 0.21 9 0.24 20 9 The Words 10 0.22 -0.54 44 1.85 2 35 of Love

References [1] Jingtian Zhang, Research on TV series hotness based on principal component analysis [J]. Technology and Economic Guide, 2018, (2). (in Chinese) [2] Chunlian Liu, A brief discussion on the dilemma of the audience rating of high-quality dramas—a case study of the Rise of Phoenixes [J]. Radio & TV Journal, 2018, (12) 42-43. (in Chinese) [3] Bing Hu, Ji Deng, Research on the influence of microblog on TV ratings [J]. Today’s Mass Media, 2015, (6). (in Chinese) [4] Conglu Gui, Eliminate the illness of “flow is greater than everything” in the industry [N]. People's Daily, 2018-10-11. (in Chinese) [5] Yong Zhou, Xuan Zhao, Construction of index system for evaluation of audio-visual communication effect in the environment of media convergence -- big data calculation and analysis based on VAR model [J]. International News, 2017, 39 (10)125-148. (in Chinese) [6] Information on http://www.guduomedia.com. [7] Information on https://index.baidu.com. [8] Information on https://www.iqiyi.com. [9] Information on https://v.qq.com. [10] Information on http://youku.com. [11] Information on https://www.douban.com.