Applied Mechanics and Materials ISSN: 1662-7482, Vols. 644-650, pp 1791-1795 doi:10.4028/www.scientific.net/AMM.644-650.1791 © 2014 Trans Tech Publications, Switzerland

Submitted: 2014-07-20 Accepted: 2014-07-21 Online: 2014-09-22

The Research in Satellite Television Channel Landing Fee of China Based on PageRank Algorithm Xin Wang1, a JianBo Liu2, b JianPing Chai3, c Hanlei Wang4, d 1,2,3,4

Information Engineering School, Communication University of China, Beijing, 100021, China

a

[email protected], [email protected], [email protected], [email protected]

Keywords: landing fee pagerank

broadcasting television stv channels

Abstract. While providing signal transmission services for stv channels (satellite television channels), catv operators in the area of broadcast and television in China raised a controversial issue about the standard of charging channel operators landing fee. To address the problem, this passage suggests a proposal about charging landing fee according to the seat value, which assessing each channel’s seat value based on the data of the audiences and the pagerank algorithm as the index. The result shows that our proposal improved the audience rating of channel while bringing a greater economic benefit to catv operators compared with the traditional standard of charging fixed landing fee. Introduction Landing fee is known as the transmission fee of the television programs collected from non-local channel operators by local catv operators. In early years, the television programs is transmitted free of charge by the approach of interdepartmental peer landing in the area of broadcast and television in China due to the limited number of stv channels. However, with the rapid increase of stv channels since 2003, an intense competition about the audience rating of channel and advertising has been raised. To get a higher share, anhui stv attempted market means and developed the mode of channels paying-landing throughout the country which established the precedent of channel operators paying for the landing fee. Because of that, other stv channels were required to pay as well, which led to the widespread of landing fee. There are mainly two kinds of standards of charging landing fee in the area of broadcast and television in China: Fixed fee of internet connection and proportional fee deducted from channels subscription similar to transaction cost. The standards and influences of landing fee raised a lot of arguments. Through establishing the Panel data model, Hanjiang Fei found positive correlations between stv landing fee and the amount of catv subscribers, the local economy, population, dialect and whether included by the csm research, while negative correlation exists between landing fee and the number of stv channels transmitted by catv operators[1]. Kui Lv analyzed economically from the theory of bilateral market suggesting that the major influences of landing fee are the price elasticity of demand, the multiple characteristics of channel operators, the single characteristics of audiences, the scale of audiences, the allowance of catv operators provided to audiences, competition between catv operators and the services differentiation [2]. This passage address a proposal about charging landing fee according to the seat value, which assessing channel seat value based on the data of the television users and pagerank algorithm as the index. Theoretical basis 1, The effects of spatial proximity The effects of spatial proximity[3] indicates the effect of the interconnection between various regional economic activities or influences of spatial relationships: negative correlation between the effects of spatial proximity and the spatial distances.

All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of Trans Tech Publications, www.ttp.net. (#69690616, Pennsylvania State University, University Park, USA-12/09/16,09:01:45)

1792

Machine Tool Technology, Mechatronics and Information Engineering

Causes of the effects of spatial proximity： (1)To require a lower social work cost, any economic activity tends to apply principle of proximity when running the business of producing and managing. (2) Limited by spatial perception, economic activities usually depend on the information from nearby territories. 2, Icebergs theory Icebergs theory originates from international economics and trade where transportation cost playing an major role. However, it would be complicated to establish an a general equilibrium model in economic geographical considering the transportation as well as the production. More over, transportation cost would break the premise of constant income elasticity of demand which simplifying Dixit-Stiglitz’s model. To avoid these, Paul Samuelson(1952) proposed a hypothesis in his international trade theory: a portion of products would be removed during transporting, which transferred cost to the products. In the new model, proportion is consumed by fixed ratio according to the distance of transportation. For example, industrial products will be consumed 1% per mile[4]. In short, Samuelson assumed that a proportion of the productions are disappeared in transit just like the melting icebergs. Only part of productions are able to get to the destination. The consumed part is the transportation cost. 3, The introduction of pagerank Pagerank, an exclusive algorithm of google, created by doctor Brin and Page in the Stanford University in 1998[5]，usually applied to measure the importance of a web page compared with other pages in the search engine. The votes of a page depend on the importance of all the links: one link one vote and the pagerank value measured through recursion based on all the pages linked to it (vbulletin). A higher rank would be given to a multi-linked web page while a non-linked one would have no rank at all. Pagerank is widely applied in the internet industry. LinCheng Jiang and Bin Ge find option leaders in the bulletin board system(bbs) by the improved pagerank algorithm[6]; Fercoq,O improved the pagerank in the spam detection[7]; ChungChi Huang and LunWei Ku use pagerank in text mining so as to detect the preferences of readers[8]. 4, The seat value of channels We propose the channel seat value based on the effects of spatial proximity, icebergs theory and pagerank. The bandwidth in the cable television network in china is around 450~550M. Since the spectral edge frequency characterized by the poor signal and strong interference is unable to transmit the program, only around 40 sets of channels can be carried by the net. As a result, besides 13 sets of cctv programs and local catv programs, only around 30 sets of non-local stv programs can be transmitted. Catv operators provide the better frequency bands for 50 sets of stv channel operators by bidding in China to improve the audience rating of channel. To get the better seat, this passage applies pagerank into the arrangement of stv channels to find the most frequently turn-into channels and provide the ground for the standard of charging landing fee. Channel seat ranking based on pagerank 1. Data structure and sampling in broadcast television system All the data used by our research comes from viewing behavior database of broadcasting television audiences, which is composed of the user background data, user rating data and program broadcasting data. Fig.1 shows the summarized data structure. The user background data provide user information, family information and information of set-top boxes. The user rating data provide the information of rating program and the information of rating user. The program broadcasting data provide channel information and program information. Three parts associate with each other, supporting the statistical analysis of channel’s seat and the standard of charging landing fee.

Applied Mechanics and Materials Vols. 644-650

1793

Database

User background data

User information

Family information

User rating data

Installed information

Rating program information

Program broadcast data

Rating user information

STV Information

Program information

Fig.1 The viewing behavior database of broadcasting television audiences We selects data from a province in china as the model analytic sample. The user background data include the background information of fifty thousand users. The user rating data include the viewing information of audiences in a month. The program broadcasting data include program information of 31 sets of paid landing channels (mostly consists of stv channel, free-of-charge channels like cctv and local tv channels are not included). Take each audiences 10 jump records each day for example, it would be 50000 users * 31 stv channels * 30 days * 10 records per day = 465 million records of stv channel switching. 2.Landing seat model of stv channel User background data

User rating data

Program broadcast data

Data Association User viewing behavior data

STV switching Record

Calculation Of STV switching matrix

STV switching Record after screening

STV switching matrix

Top Ranking of STV seats

Fig. 2 Block diagram of landing seat model of stv channel Fig.2 shows the block diagram of landing seat model of stv channel: (1) Obtain user viewing behavior data: associating the user background data, user rating data and program broadcasting data to get the user viewing behavior data. (2) Calculation of stv switching matrix  Data statistics and analysis：calculating stv switching record through data statistics and analysis of user viewing behavior data.  Data filtering and screening：selecting 31 sets of paid landing stv switching record by filtering and screening all stv switching record(mostly stv channel, free-of-charge channels like cctv and local tv channels are not included).  Data fusion and recombination: fusing and recombining the filtered data as stv switching matrix. (3) Calculate the top ranking of channel seat: calculating seat score and the top ranking of stv channel through pagerank algorithm. 3. Stv channel seat ranking based on pagerank This passage calculates the frequently turn-into channels’ seat using pagerank algorithm. Pagerank algorithm formula: ( )

()

()

(1)

1794

Machine Tool Technology, Mechatronics and Information Engineering

Assuming that stv i, j, ..n are possible to turn to stv a, then pr(a) represents the pagerank value of stv a; c(a) represents the turning times from stv a to other stv channels. In the general case, the original pagerank value of each stv channel is 1. Then applying the formula in calculating pagerank value of each channel through recursive computation until the values tend to be stable. Because of the randomized design of inter channel switching in audience behavior model, the value of each stv pagerank would be divided equally to all stv channels pointing to this channel, where pr(i)/c(i) in the formula shows the strategy of equal division. Experimental results and analysis (1)Obtain the audience behavior data like Table 1 Table1 Audience behavior data cardnum date starttime endtime 3890898004 2013/9/8 8:05:12 10:18:47 3890898004 2013/9/8 11:02:59 11:03:16 3890898004 2013/9/8 11:03:17 11:03:33 (2)Analyze the switching matrix of stv channel

time 8015(second) 2014(second) 316(second)

stv gansu stv anhui stv cctv-1

Fig.3 Visualized figure of channel switching matrix Fig.3 illustrates the visualized figure of channel switching matrix. Word indicates the channel names; The size of word represents the synthetical importance of inter channel switching; the direction of segment indicates the direction of switching; the line weight represents the frequency of inter channel switching. It is obvious that channel seat like zhejiang stv, jiangsu stv, shenzhen stv, guangzhou stv and tianjin stv are frequently arrived. (3) Calculate the top rank of stv channel seat

Fig.4 Seat value of stv channel

Applied Mechanics and Materials Vols. 644-650

1795

Fig.4 shows the seat value of each stv channel. x-coordinate gives the names of stv channel while y-coordinate gives the seat value of stv channel calculated through pagerank. Zhejiang stv, jiangsu stv, shenzhen stv, guangzhou stv and tianjin stv occupy the top 5, which responding the analysis in the visualized figure of channel switching matrix. Located in the most frequently arrived seat, these seats and surrounding seats should be charged higher landing fee by catv operators since the programs of these seats are most likely to attract audience’s attention.. Summary Catv operators require stv operators to pay landing fee while providing the services of signal transmission. It remains a controversial problem setting standard of charging landing fee. This passage proposes the channel seat theory based on economic theory of the effects of spatial proximity and icebergs theory. Supported by a whole month switching data of fifty audiences in a province in china and the programs of 31 paid landing stv, our seat theory finds the golden seat to help charging landing fee based on pagerank algorithm. However, since the switching connection between channels would change with some factors, such as the preference of audience and the channel viewing rate. The result of the traditional pagerank algorithm can only be applied to a limited period (per season or per month, for instance) when catv operators deduct a certain percentage from the incomes of stv channels. In the next stage of research, we will study golden seat based on the time series and combine the multiple indexes like channel viewing rate, channel switching rate and channel satisfaction degree to provide a more dependable indication of charging stv landing fee. References [1] Haijiang Fei, Empirical Research on Satellite Television Landing Fee Based on Panel Data Model[J]Finance and Accounting, 4th periodical of 2012 [2] Kui Lv. Cable Television Network Access Fee Pricing Model and Influence Factors Analysis Based on Two-sided Market Theory[J]College Journal of Xidian University Social Science Edition , 1st periodical of 2009 [3] Dicken, P. and A. Malmberg. "Firms in territories: A relational perspective." Economic Geography 77(4): 345-363,2001 [4] Paul Krugman."Increasing Returns and Economic Geography."Journal of Political Economy 99(3):483-499, 1991. [5] PageL, BrinS, MotwaniR, etal. The PageRank citation ranking Bringing order to the web[C]//Stanford Digital Libraries Working Paper,1998. [6] Lincheng Jiang, Bin Ge, Weidong Xiao, etal. “BBS opinion leader mining based on an improved PageRank algorithm using MapReduce.” Chinese Automation Congress (CAC), 2013 [7] Fercoq,O. “PageRank optimization applied to spam detection.” Network Games, Control and Optimization (NetGCooP), 2012 [8] Chung-Chi Huang, Lun-Wei Ku. “Interest Analysis Using Semantic PageRank and Social Interaction Content.” Data Mining Workshops (ICDMW), 2013