Provided by the author(s) and NUI Galway in accordance with publisher policies. Please cite the published version when available. Using social media for online television adaptation services at Title RTÉ Ireland Barraza-Urbina, Andrea; Hromic, Hugo; Heitmann, Benjamin; Author(s) Tamatam, Himasagar; Yañez, Andrea; Hayes, Conor Publication 2016 Date Barraza-Urbina, Andrea, Hromic, Hugo, Heitmann, Benjamin, Tamatan, Himasagar, Yañez, Andrea, & Hayes, Conor. (2016). Publication Using social media for online television adaptation services at Information RTÉ Ireland (pp. 91). Galway: Insight Centre for Data Analytics, National University of Ireland, Galway. Insight Centre for Data Analytics, National University of Publisher Ireland, Galway Link to publisher's http://dx.doi.org/10.13025/S8059M version Item record http://hdl.handle.net/10379/5709 DOI http://dx.doi.org/10.13025/S8059M Downloaded 2023-01-01T01:08:02Z Some rights reserved. For more information, please see the item record link above. Using Social Media Data for Online Television Adaptation Services at RTÉ Ireland Final Project Report ADDRESSED TO: Raidió Teilifís Éireann AUTHORS: Andrea Barraza-Urbina Hugo Hromic Benjamin Heitmann Himasagar Tamatam Andrea Yañez Conor Hayes INSIGHT CENTRE FOR DATA ANALYTICS National University of Ireland, Galway, Ireland Unit for Information Mining and Retrieval (UIMR) GALWAY, IRELAND 2016 i FOREWORD RTÉ (Raidió Teilifís Éireann) is the national provider of Television (TV) and radio in Ireland. RTÉ broadcasts its content online through the RTÉ Player and provides services to interact with its users using social media, such as Twitter and Facebook. However, RTÉ wishes to exploit the full power of knowledge that can be obtained from social media, and with that knowledge enhance their online services to further engage users. Towards this goal we present the RTÉ XPLORER prototype. This prototype offers services based on both Social Analytics and Information Adaptation. The overall goal is to offer services to RTÉ end users to support them in exploring the RTÉ product catalogue and understanding what is happening in social media related to RTÉ programming. In this manner, users could find interesting content faster and be encouraged to participate in social media communities discussing RTÉ content. The goal of this document is to present the project outcomes, the process to achieve these outcomes and conclusions from our social analytic studies. In the document, we first present the concepts that laid foundation to our research and related work. Next, we review the data that characterizes the RTÉ case study, including data sources and data collection strategies. We proceed towards presenting the services we propose for both the end-users of the RTÉ Player service and directives at RTÉ. Finally, we present the implementation of a functional prototype meant to represent how proposed services could integrate with the current RTÉ Player service. ii TABLE OF CONTENTS 1 INTRODUCTION ................................................................................................. 1 1.1 MAIN DELIVERABLES ....................................................................................... 1 1.2 RESTRICTIONS AND KEY CHALLENGES ............................................................. 1 1.3 DOCUMENT OUTLINE ...................................................................................... 2 2 BACKGROUND AND RELATED WORK ........................................................... 5 2.1 TV, SOCIAL CURIOSITY AND SOCIAL MEDIA ...................................................... 5 2.2 TWITTER: THE GRAPEVINE FOR TV FANS ......................................................... 6 2.3 INFORMATION ADAPTATION AND RECOMMENDER SYSTEMS FOR TV SERVICES .... 7 3 REVIEW OF RTÉ DATA ..................................................................................... 9 3.1 USE CASE DATA OVERVIEW ............................................................................ 9 3.1.1 Domain Model ................................................................................................... 9 3.1.2 Adaptation Model ............................................................................................. 12 3.1.2.1 User Profile ............................................................................................................. 12 3.1.2.2 Contextual Profile ................................................................................................... 15 3.2 DATA SOURCES ........................................................................................... 17 3.2.1 RTÉ Player Service ......................................................................................... 17 3.2.2 Social Media Platforms .................................................................................... 18 3.2.3 Online Databases ............................................................................................ 18 3.3 DATA ACCESS RESTRICTIONS ....................................................................... 19 3.4 DATA COLLECTION STRATEGIES .................................................................... 20 3.4.1 Web Page Crawler ........................................................................................... 20 3.4.2 Twitter Data Collection ..................................................................................... 21 3.4.2.1 Data Selection: What to listen to from Twitter? ...................................................... 22 3.4.2.2 Data Cleaning ......................................................................................................... 24 3.5 DATA INTEGRATION STRATEGY ...................................................................... 25 3.6 SUMMARY .................................................................................................... 25 4 SOCIAL ANALYTICS ON TWITTER ................................................................ 26 4.1 BASIC DATA ANALYSIS METHODOLOGY .......................................................... 26 4.2 DISCUSSION ON USER BEHAVIOURS RELATED TO RTÉ SHOWS ......................... 27 4.2.1 Descriptive Statistics ........................................................................................ 28 4.2.2 How Do Users Tweet about RTÉ Programmes? .............................................. 30 iii 4.2.3 Do Users Engage in Conversations in Twitter? ................................................ 34 4.3 COMMUNITY DATA ANALYSIS METHODOLOGY ................................................. 37 4.3.1 Data Processing Windows ............................................................................... 39 4.3.2 Community Detection....................................................................................... 39 4.3.3 Community Tracking ........................................................................................ 41 4.4 DISCUSSION OF RTÉ PROGRAMMES IN COMMUNITIES ..................................... 42 4.4.1 Community Statistics ....................................................................................... 43 4.4.2 How Do Users Gather in Communities About RTÉ Programmes? ................... 44 4.5 RTÉ PROGRAMMES CO-OCCURRENCE IN TWITTER ......................................... 47 4.5.1 Tweet-based Co-occurrence ............................................................................ 47 4.5.2 Community-based Co-occurrence ................................................................... 49 4.6 SUMMARY .................................................................................................... 50 5 SOCIALENS FOR RTÉ .................................................................................... 52 6 RTÉ XPLORER PROTOTYPE SERVICES ...................................................... 55 6.1 RTÉ XPLORER PROTOTYPE LOGICAL ARCHITECTURE ..................................... 55 6.2 PRESENTATION ADAPTATION ......................................................................... 57 6.3 CONTENT ADAPTATION ................................................................................. 62 6.3.1 Community Analytics ....................................................................................... 62 6.3.2 Recommendation Engine ................................................................................. 64 7 RTÉ XPLORER PROTOTYPE IMPLEMENTATION ........................................ 68 7.1 PROTOTYPE IMPLEMENTATION AND DEPLOYMENT ........................................... 68 7.1.1 Frameworks and Technologies ........................................................................ 68 7.1.2 Data Model Implementation ............................................................................. 70 7.1.3 Deployment Requirements .............................................................................. 72 7.2 PROTOTYPE USER INTERFACE ....................................................................... 73 7.2.1 Exploration View .............................................................................................. 73 7.2.2 Video View ....................................................................................................... 77 7.3 SUMMARY .................................................................................................... 79 8 CONCLUSION .................................................................................................. 81 8.1 CONCLUSION ............................................................................................... 81 8.2 FUTURE WORK ............................................................................................. 84 9 REFERENCES.................................................................................................. 87 iv TABLE OF FIGURES FIGURE 1. ADAPTATION AND ANALYTICS SERVICES FOR RTÉ ........................................... 3 FIGURE 2. DOMAIN PROFILES ...................................................................................... 11 FIGURE 3. USER PROFILE ............................................................................................ 14 FIGURE 4. CONTEXTUAL PROFILE ................................................................................ 16 FIGURE 5. OVERVIEW OF TWITTER DATA AND RTÉ PROGRAMMING DATA INTEGRATION. ... 26 FIGURE 6. OVERVIEW OF THE TWITTER POSTING ACTIVITY IN THE CAPTURED DATA. .......... 28 FIGURE 7. DISTRIBUTION OF RECEIVED TWEETS TO PROGRAMMES FOR ALL USERS (LEFT) AND ONLY POSTED BY THE OFFICIAL RTÉ ACCOUNTS (RIGHT) CONFIGURED FOR LISTENING. .......................................................................................................... 31 FIGURE 8. LOG-SCALE HISTOGRAMS OF THE NUMBER OF PROGRAMMES IN THE SAME TWEET (LEFT) AND BY THE SAME USER (RIGHT). ................................................................ 33 FIGURE 9. HISTOGRAM OF USER RECIPROCITIES BASED ON REPLIES (LEFT) AND RETWEETS (RIGHT). .............................................................................................................. 35 FIGURE 10. HISTOGRAM OF THE USER POPULARITIES BASED ON RETWEETS. ................... 35 FIGURE 11. HISTOGRAM OF USERS OVERLAP FOR RETWEET POPULARITY AGAINST REPLY (LEFT) AND RETWEET (RIGHT) RECIPROCITIES. ...................................................... 36 FIGURE 12. HISTOGRAM OF USERS OVERLAP FOR REPLY AGAINST RETWEET RECIPROCITIES ................................................................................................... 37 FIGURE 13. TWITTER USER-USER DIRECTED WEIGHTED NETWORK. ................................ 40 FIGURE 14. (LEFT) SAMPLE COMMUNITIES (BLUE) IDENTIFIED AROUND RTÉ PROGRAMMES (YELLOW) AND HASHTAGS (RED) DURING AN HOUR OF TWEETS CAPTURING IN JULY, 2015. A TOTAL OF 36,648 USERS AND 23,232 TWEETS ARE USED. (RIGHT) CLOSE-UP VIEW OF A SAMPLE COMMUNITY INVOLVING THE EASTENDERS, CASUALTY, HOLBY CITY AND NEIGHBOURS SHOWS. ................................................................................... 41 FIGURE 15. COMMUNITY EVOLUTION OVER TIME FOR DISCUSSIONS ABOUT RTÉ PROGRAMMES. .................................................................................................... 42 FIGURE 16. HISTOGRAMS FOR NUMBER OF COMMUNITIES IN WINDOWS (LEFT) AND LOG- SCALED NUMBER OF USERS IN COMMUNITIES (RIGHT). ............................................ 43 FIGURE 17. DISTRIBUTION OF THE NUMBER OF COMMUNITIES FOUND PER RTÉ PROGRAMME .......................................................................................................................... 44 FIGURE 18. TWEET (RED) VS RETWEET (BLUE) ANNOTATIONS ........................................ 46 v FIGURE 19. MATRICES SPARSITIES OVER WEEKLY PERIODS FOR THE DIFFERENT PERSPECTIVES OF CO-OCCURRENCE OF PROGRAMMES (TWEETS-BASED AND COMMUNITY-BASED). ........................................................................................... 49 FIGURE 20. COMMUNITY ROLES WIDGET IN SOCIALENS ................................................ 54 FIGURE 21. EXAMPLE WIDGETS IN SOCIALENS ............................................................. 54 FIGURE 22. RTÉ XPLORER PROTOTYPE LOGICAL ARCHITECTURE .................................. 56 FIGURE 23. IMPLEMENTATION FRAMEWORKS AND TECHNOLOGIES .................................. 69 FIGURE 24. EXPLORATION VIEW OF RTÉ XPLORER PROTOTYPE .................................... 74 FIGURE 25. “JUST FOR YOU” SECTION OF THE MAIN LANDING PAGE ................................. 75 FIGURE 26. “TODAY IN SOCIAL MEDIA” SECTION OF THE MAIN LANDING PAGE ................... 76 FIGURE 27. “MORE RTÉ CONTENT” SECTION OF THE MAIN LANDING PAGE ....................... 77 FIGURE 28. VIDEO VIEW OF RTÉ XPLORER PROTOTYPE ............................................... 78 vi TABLE OF TABLES TABLE 1. TOP 10 COUNTRIES IDENTIFIED IN PLACES DATA AND THEIR NUMBER OF TWEETS 29 TABLE 2. TOP 10 PROGRAMMES AND THEIR NUMBER OF MENTIONING TWEETS. ................ 31 TABLE 3. RTÉ XPLORER PROTOTYPE EXPLORATION VIEWS ........................................... 61 TABLE 4. RTÉ XPLORER PROTOTYPE COMMUNITY ANALYTICS WIDGETS ........................ 64 TABLE 5. TYPES OF PROGRAMME RELEVANCE ............................................................... 66 TABLE 6. MONGODB COLLECTIONS ............................................................................. 71 vii 1 I NTRODUCTION RTÉ (Raidió Teilifís Éireann) is the national provider of Television (TV) and radio in Ireland. RTÉ broadcasts its content online through the RTÉ Player and provides services to interact with its users using social media, such as Twitter and Facebook. However, RTÉ wishes to exploit the full power of knowledge that can be obtained from social media, and with that knowledge enhance their online services to further engage users. For this goal, RTÉ joined forces with The Insight Centre for Data Analytics. This document outlines the project outcomes of this collaboration. In this section, we first outline the main deliverables of the project. Next, we highlight the restrictions and key challenges faced. Finally, we present the document outline. 1.1 Main Deliverables The three main deliverables are:  RTÉ XPLORER Prototype: The demo offers services based on both Social Analytics and Information Adaptation approaches. The RTÉ XPLORER prototype is meant to be a tangible representation of how services proposed in this document could be integrated in to the RTÉ Player service.  Publication at International Conference: Barraza-Urbina, A., Hromic, H., Hulpus, I., Heitmann, B., Hayes, C., Cantle, N. Using Social Media Data for Online Television Recommendation Services at RTÉ Ireland. 2nd Workshop on Recommendation Systems for Television and Online Video, 9th ACM Conference on Recommender Systems, 20/09/2015.  Project Outcomes Document: Barraza-Urbina, A., Hromic, H., Heitmann, B., Tamtam, H., Hayes, C., Using Social Media Data for Online Television Adaptation Services at RTÉ Ireland. 1.2 Restrictions and Key Challenges The main research barrier was on the amount and quality of the data that we had access to, concerning RTÉ. 1 On the one hand, in terms of user preference data related to programmes, it is understandable that RTÉ cannot freely share user profile data. In addition, given that RTÉ users are not obligated to sign-in to use the RTÉ Player service, RTÉ has explained that most of their users do not have user accounts. This means that if they could share data freely, nevertheless the amount of data might not be enough to offer quality Information Adaptation services. On the other hand, in terms of programme-related data, RTÉ's programme catalogue is highly dynamic, as they add programmes and remove them on a daily fashion. As a consequence, in order to offer services tailored to the user's unique characteristics, we face the following challenges:  Lack of personal preference data such as ratings.  Relatively little historical user session information.  Dynamic inventory and limited life span of recommendable items.  No integration of social media analytics.  Users would be considered anonymous. We address these challenges by identifying a key opportunity in using data immersed within social media as a valuable resource that can be exploited to better understand user show preferences. Consequently, Insight has addressed the endeavour of providing a set of solutions based on social media that resulted in analytics tools for decision makers and Information Adaptation services to enhance the RTÉ Player service. The goal of this document is to offer a detailed presentation of the proposed solutions. 1.3 Document Outline This section offers an outline of the document structure. The goal of this document is to present the project outcomes, the process to achieve these outcomes and conclusions from our social analytics studies. We have focused on providing RTÉ with a solution that offers two types of services, Social Analytics and Information Adaption services. These services are intended for two types of end-users: the directives/employees at RTÉ and end-users of the RTÉ Player service. The services and target users can be viewed in Figure 1. 2