Social Media Analytics Challenges and Case Studies
- 1. Social Media Analytics Challenge and Case Studies Big Data Engineering, DPU
- 2. Global Usage of Social Media Analytics Big Data Engineering, DPU
- 3. PNERN ASAVAVIPAS (Nern) CEO & Co-Founder @ OBVOC Education - B.A. Economics, Thammasat University Experiences - 6 years in Social Listening & Analytics
- 4. What is Social Media Analytics?
- 5. Social Listening Social Analytics Social Intelligence Capturing words Analyzing Data & Discovering Insights Actioning Insights
- 6. Nespresso Using Social Insights to Build a Global Marketing Strategy
- 7. Consumer Insights from Co鍖ee Drinking Discussion on Social Source: Crimson Hexagon
- 8. Customer Insights through Audience Competitive Analysis Source: Crimson Hexagon
- 10. How data contribute Marketing Strategy Source: Crimson Hexagon Commercial Advertising Events and Pop Up Store Sponsorship
- 11. Commercial Advertising Source: Crimson Hexagon Penelope Cruz - Actress Lana Del Rey - Singer
- 12. Penelope Cruz - Nespresso TV Commercial, Song by Lana Del Rey
- 13. Developing Marketing Strategy Based on Insights
- 14. Developing Marketing Strategy Based on Insights
- 15. Developing Marketing Strategy Based on Insights
- 16. Nespresso Monitoring & Learning From Marketing Efforts
- 17. Measuring the E鍖ect of Marketing E鍖orts Brand Perception: Advocacy Intent to Purchase Brand Perception: Positive Reputation Brand Perception: Negative Reputation Marketing E鍖orts Source: Crimson Hexagon
- 18. Measuring Consumer Exposure to the Brand Opening of Nespresso boutique store in San Francisco Airing of commercial featuring Penelope Cruz and Lana Del Rey Beginning of Pop Up events at Americas Cup, SXSW, and cities throughout the US Source: Crimson Hexagon
- 19. Measuring Consumer Exposure to the Brand Source: Crimson Hexagon
- 20. Measuring Consumer Exposure to the Brand Source: Crimson Hexagon
- 21. Identifying Key Areas of Success Source: Crimson Hexagon
- 22. Driving Consumer Purchase Intent The pop-up events increasing from 3% to 21% over the same time period. Source: Crimson Hexagon
- 23. Key Takeaway Source: Crimson Hexagon With actionable insights, Nespresso could learn from a targeted campaign, monitor e鍖orts to learn what marketing events worked and engaged their target audience, and build a global strategy for the future. As with Nespresso, social insights enable any brand or agency to: 1. Know where to reach new and relevant customer segments and what messaging will most e鍖ectively engage them. 2. Track speci鍖c campaign e鍖orts and customer interaction with them, to know what worked. 3. Categorize and measure customer interaction to learn about brand perception, method of exposure, and future opportunities for engagement. 4. Identify key areas of success and quantify return on investment through changes in intent to purchase and competitor benchmarking.
- 30. Process & Implementation Big Data Engineering, DPU
- 31. PUTTASAK TANTISUTTIVET (TOR) Senior Data Research Analyst @ THOTH ZOCIAL Education - B.Sc. Electronics Engineering, KMITL - M.Sc. Business Administration (Marketing) , NIDA Experience - Social media strategy - Digital Project Planner - Content Strategy - Data Research and Analyst - Speaker @Young Data Scientist, J.Walter Thomson, Nitade Chula
- 33. Data Team Data Team = Chef
- 34. Chef Raw Material Cooking Decorate and Serve
- 35. Data Team Data Collecting (Raw Material) Data Processing (Cooking) Delivering Insight (Decorate and Serve) - NLP - Text Mining - Categorize - Cleansing Data - Association Rule - Correlation - Classification - Clustering - Timeline Analysis - Data Warehouse - Database - Sourcing - Social Media - Data Hacking - Tools - Presentation - Paper - Research
- 36. Social Media Data Process Delivering InsightData Collecting Data Processing
- 37. Social Media Data Process Delivering InsightData Collecting Data Processing
- 38. 2.5 Billion Talks On 2016 ~ 5 Million Talks every day ~ 5,000 Talks every minute ~ 80 Talks every second
- 39. Data Collecting. - Millions Millions message per month (and Very fluctuate) - Depend on platform, Always Change! - Junk Source and Data - Data Limitations - Both Structure and Unstructure Data - Varieties type of Data (Text, Video, Image, Numbers) - Coverage VS Accuracy
- 40. Social Media Data Process Data Collecting Data Processing Delivering Insight
- 41. Data Processing. - Natural language processing - Text Mining - Cleansing Data - Categorize - Association Rule - Correlation - Classification - Clustering - Data Modeling
- 42. NLP Procedure Data (input) Word segmentation Feature extraction Weight optimization Results (output) Classification Word Sentence Paragraph Document Tokenization Part-of-speech tagging Binary Numeric Bag-of-words (BOWs) TF-IDF Sub-words (prefixes, suffixes) etc. Supervised learning Machine learning Unsupervised learning Clustering Target output Sentiment score Opinion Keywords etc.
- 43. Social Media Data Process Delivering InsightData Collecting Data Processing
- 44. Data interpretation. 犖犖ム険犖犖犖迦犖犖犖犖犖犖犖ム険犖犖÷顕 犖犖項犢犖犖犖劇犖犖犖犖巌犖犖犖犖犖伍犖犖項犖∇鹸犖 犖犖犖伍犢犖ム鍵犖÷元犖犖犖犖犖巌犖犖犖犖犖ム険犖犖÷顕犢犖犖犖犖謹犖犖犖犖謹犖 犖犖犖 犖犖犖犖犖犖犖犖園犖犖犖犢犖犖犖伍犖犖萎犖犖犖巌検犢犖犖犖迦鍵犢犖犖犖園犖犖о犢犖 犖犖犖犢犖犖犖劇犖犖犖犖巌? 犖 ) 犖犖о犖犖朽犖犖項犖∇鹸犖犢犖∇賢犖 犖) 犖犖о犖犖朽犢犖￥犖犖項犖∇鹸犖
- 45. Turn Data into Value Information
- 47. - Text - Image - Video - Time - Location - Engagement - Etc. - Segmentation - Interested - Behaviors - Trend - Attitude / Perceptions - Unknown Insight Data interpretation.
- 50. Challenge: Social Media Data Big Data Engineering, DPU
- 51. Social Media Data is Junk VS Accuracy
- 52. Crawled Data iPhone 7S iPhone7S iPhone7 S iPhone 7S 犢犖犢犖犖7s 犢犖犢犖犖 7s 犢犖犢犖犖 7s 犢犖犢犖犖7 s 犢犖犢犖犖7s 犢犖犢犖犖 7s . . . Huawei P9 Plus 5*2*3*2*2 = 120 Keywords Subject (Huawei | 犖犖園硯犢犖犖э犖≒ | 犖犖園硯犢犖э犖≒ | 犖犖園硯犢犖犖э犖≒) & (P9plus | P9 plus | P9+ | 犖犖9)
- 53. Social Media Data is World of ASSUMPTION
- 54. Mobile Voice | Assumptions - Voice 犖犖犖犖ム犖犖萎犖犖犖犖犖о顕犖÷賢犖∇顕犖犖犖劇犖犖犖犖 Users 犢犖犖? - 犖犖犖犖犖迦犢犖犖ム鍵犖犖犖犖犖迦 犖÷元犖犖ム犖犖萎犖犢犖犖迦犖園犢犖犖? - 犢犖犖迦犖犖犖迦鍵 Users 犢犖犖犖 犖犖犖劇賢犢犖犖迦犖朽 Page 犢犖犖犖犖犖о権? - 犖犖犖犖迦権犖犖項犖犖謹犖 IG 犢犖ワ犖о犖 #Huaweip9 犖犖萎犖園犢犖犖? - 犖犖迦牽犖犖朽犖犖犖犖項犖э犖迦犖朽犢犖ム犖園肩犖犖朽犖迦権 = 犢犖犖萎犢犖迦犖犢犖犖犖劇犖犢犖犖÷見犖犖劇賢犢犖犖犖犖? - 1 犖犖犖犖о顕犖÷検犖朽犖迦牽犖犖項犖犖謹 2 犖犖醐犖 犖犖萎犖園犢犖∇犖犖園犖犖犖劇賢犖犖園犖犖о検 - 1 犖犖犖犖о顕犖÷犖朽犖犖項犖犖謹 Feature 犖犖迦犢 犖犖萎犖園犢犖∇犖犖犖劇賢犖犖園犖犖о検 - 1 犖犖犖犖о顕犖÷犖朽犢犖犖犖園 Engagement 犖犖萎犖犖ム犖о顕犖÷見犖÷顕犖∇犖犖 Engagement 犖∇険犖犢犖犖犖?
- 55. Social Media Data is Thai Culture and Lang is not easy
- 56. Beautiful of Thai Lang
- 57. Beautiful of Thai Lang
- 59. Social Media Data is Not fit for all
- 61. Social TV Rating - Data 犖犖伍犢犖犖犖犖 Social Media 犖犖朽犢犖犖犖犖園硯犢犖犖犖犖犖 Rating? - 犖犖 Video View? - 犖犖 Voice 犖犖朽犖犖項犖犖謹犖犖迦権犖犖迦牽犖犖園犖? - 犖犖 Engagement 犖犖 Video? - 犖犖園犖犖о顕犖÷肩犢犖迦犖園犖犖萎見犖э犖迦犖犖犖÷弦犖ム犖犖 Live Video 犖犖園 VOD 犖∇険犖犢犖? - 犖犖萎犖犖犖 FB Video 犖犖犖劇賢 Youtube Video 犖犖犖劇賢犖犖о検犖犖園? - Facebook Live 犖犖劇賢 Time Base 犖犖犖劇賢 VOD? - 犖犖萎犖犖 Voice 犢犖犖 Rating 犖犖犖劇賢犢犖犖 Amplifier? - 犢犖ワ犖 Sentiment 犖犖ム鍵?? - 犖犖迦犖迦犖犖迦犖￥犖犖迦検犖迦牽犖犢犖犢犖犢犖
- 62. Social TV Rating 3 Sec = 1 Views 30 Sec = 1 Views
- 63. Challenge: Business Stakeholder Big Data Engineering, DPU
- 67. Jargon!
- 69. How to prevent?
- 71. Start with small and Build Trust
- 73. 1. Engage with higher position as much as you can. 2. Start with small and build their trust. 3. Be careful your jargon words. 4. Keep learning on your fields. 5. Talk less and show more. Key Takeaway
- 74. Enjoy Analyzing!
- 75. Case Studies Big Data Engineering, DPU
- 76. Ways to apply Social Media Data 犖犖÷犖 犖犖 犖犖迦犖 犢犖犖犖犖 Use Correlation Technique MOMs Insight
- 77. Ways to apply Social Media Data Answer : 33% 犖犖劇犖, 31% 犖犖, 21% 犢犖犖巌, 21% 犖犖巌 Store Visit #1 : { 犢犖 + x ... store } (Use Association Technique) Question : 犖ム弦犖犖犖迦犖犖犢犖迦賢犖萎犖犖犖朽犖ｏ犖迦? Solution : Listen & Extract only talks about store or check-ins at store. Identify 犢犖 and adjacent keyword.
- 78. Ways to apply Social Media Data Store Visit #2 : { 犢犖 + 犖犖劇犖 + x ... store } Question : 犖ム弦犖犖犖迦犖 犖犖劇犖犖犖萎犖犖犖朽犖ｏ犖迦? Solution : Listen & Extract only talks about store or check-ins at store. Identify 犢犖+犖犖劇犖 and adjacent keyword. Answer : 12% 犢犖犖, 10% 犢犖, 77%
- 79. Ways to apply Social Media Data Store Visit #2 : { 犢犖 + 犖犖劇犖 + x ... store } Question : 犖ム弦犖犖犖迦犖 犖犖劇犖犖犖萎犖犖犖朽犖ｏ犖迦? Solution : Listen & Extract only talks about store or check-ins at store. Identify 犢犖+犖犖劇犖 and adjacent keyword. Answer : 12% 犢犖犖, 10% 犢犖, 77% 犢犖￥
- 80. Ways to apply Social Media Data 犢犖犖 犖犖о顕犖 犖犖ム賢犖犖犖園権 Condo Insight
- 81. Ways to apply Social Media Data 犢犖犖 犖犖о顕犖 犖犖ム賢犖犖犖園権犢犖÷元犖Condo Insight
- 82. Ways to apply Social Media Data 犖犖犖迦犖迦 XXX Followed by Superstar name
- 83. Ways to apply Social Media Data
- 84. Ways to apply Social Media Data 1 2 Kumamoto > Fukuoka > Nagasaki Fukuoka > Kumamoto > Mountain Aso 1 2 Mountain Aso > Kumamoto > Fukuoka Fukuoka Kumamoto City (Castle, Suizenji, etc.) Travel Behavior
- 85. Ways to apply Social Media Data Event Analysis
- 86. Ways to apply Social Media Data Event Analysis
- 87. Ways to apply Social Media Data Precisions Ads
- 88. Ways to apply Social Media Data Traditional Ads Targetings Ads Personalized Ads Improve way to advertising
- 89. Ways to apply Social Media Data CHATBOT Application and interface (Input) Natural language processing (NLP) Intelligent search BOT knowledge Keyowords/Pattern matching and sorting Ontology PlatformDictionary Conversational AI
- 90. Ways to apply Social Media Data Sarcasm Prediction
- 91. 犖犖迦犢犖э犢犖犖犖巌肩犖. - 犖犖巌犖犖犖朽犖犖∇顕犖犖犖朽犖犖伍犖犖犖犖犖迦牽犖犢犖迦犖迦犖犖犖÷弦犖ム犖劇賢犖犖迦牽犢犖￥犖÷元犖犖犖÷弦犖 - 犖犖迦牽犖犢犖迦犖迦犖犖園 Data 犖犖犖犢犖犢犖о献犖, 犖犖犢犖犖巌犖! - 犖犖迦犖犖園犖犖萎犢犖迦犖迦犖犖迦犖犖朽犖犖犖巌 犖犖犖迦牽犖園犢犖犖朽権犖犖犖迦牽犢犖犖 犖犖迦牽犢犖犖犖犖犖÷弦犖モ - 犢犖￥犢犖犖犖伍犖犖犖犖朽犖犖迦検犖迦牽犖犖犖園犖犖迦犖園硯犢犖犖犖犖犢犖犖犢犖犖犖伍犖犖迦 - 犢犖￥犢犖犖犖伍犖犖犖犖朽犖犖萎犖犖迦犖犢犖ム鍵犖犖犖 Data - Data 犢犖￥犢犖犢犖о犖犖÷犖 犢犖犢犖犖 犖犖 犖犖迦犖犖迦犖犖朽犢犖 Data 犢犖犖劇犖犖犖ｏ犖迦犢犖о犖犖÷犖
- 93. THANK!