Meenakshi Nagarajan, Kamal Baid, Amit Sheth and Shaojun Wang KNOESIS, Wright State University M. Nagarajan, K. Baid, A. P. Sheth, and S. Wang, "Monetizing User Activity on Social Networks - Challenges and Experiences“, 2009 IEEE/WIC/ACM International Conference on Web Intelligence, Milan, Italy

 Content-based advertisements (CBAs)  Well-known monetization model on the Web but not translating well on SNSs  Monetizing content on Web 2.0  Where to monetize  What to monetize  It’s the talk of the town!

May 30,June

June 01, 2009

 Interests stated on user home/profile pages do not translate to purchase intents  Interests are often outdated..  Intents are rarely stated on a profile..  Some highly demographic targeted cases work  Overall, click through stats are staggeringly low – show some

June 01, 2009 Concert tickets MP3 downloads Services in and around location

June 01, 2009

 Informal, casual nature of content ▪ People are sharing experiences and events ▪ Main message overloaded with off topic content  Non-policed content ▪ Brand image, Unfavorable sentiments 1  People are there to network ▪ User attention to ads is not guaranteed I NEED HELP WITH SONY VEGAS PRO 8!! Ugh and i have a video project due tomorrow for merrill lynch :(( all i need to do is simple: Extract several scenes from a clip, insert captions, transitions and thats it. really. omgg i cant figure out anything!! help!! and i got food poisoning from eggs. its not fun. Pleasssse, help? :( 1 Learning from Multi-topic Web Documents for Contextual Advertisement, Zhang, Y., Surendran, A. C., Platt, J. C., and Narasimhan, M., KDD 2008

 System that generates ads based on activity (user generated content) elsewhere by 1. Identifying monetizable posts: intents behind user posts  Pull content with monetization potential 2. Identifying keywords for advertizing from monetizable posts  Dealing with off-topic chatter

 User studies  Hard to compare activity based ads to s.o.t.a  So we evaluate subgoals  How well are we able to identify monetizable posts (component 1)  How targeted are ads generated using our keywords vs. entire user generated content (component 2)

Identification, Evaluation

 Scribe Intent not same as Web Search Intent 1  People write sentences, not keywords or phrases  Presence of a keyword does not imply navigational / transactional intents  ‘am thinking of getting X’ (transactional)  ‘i like my new X’ (information sharing)  ‘what do you think about X’ (information seeking) 1 B. J. Jansen, D. L. Booth, and A. Spink, “Determining the informational, navigational, and transactional intent of web queries,” Inf. Process. Manage., vol. 44, no. 3, 2008.

 Action patterns surrounding an entity (X)  How questions are asked and not topic words that indicate what the question is about  “where can I find a chotto psp cam”  User post also has an entity

MySpace User Posts (not annotated for intent) Extract all 4-grams > freq 3 Using seed words (who, when, why, what, how) Extract all 4-grams containing seed words Candidate / Potential set of patterns (S c ) ‘does anyone know how’, ‘where do i find’, ‘someone tell me where’…

10 manually picked Information Seeking Patterns S is ‘how cool are we’ is not Information Seeking Remaining candidate patterns S c = S c - S is Candidate patterns S c Goal: Evaluate candidate patterns and judge if it is Information Seeking or not ‘does anyone know how’, ‘where do i find’, ‘someone tell me where’…

‘.* anyone know how’ ‘does.* know how’ ‘does anyone.* how’ ‘does anyone know.*’ For each filler Look for patterns in candidate pool S c -Functional compatibility of filler -words used in similar semantic contexts - Empirical support for filler ‘does anyone know how’ For every known Information Seeking pattern in S is generate set of filler patterns

 Known Information Seeking patterns S is = {‘does anyone know how’, ‘where do I find’, ‘someone tell me where’}  p is from S is = `does anyone know how’  Match ‘does * know how’ with patterns in the Candidate Pool  ‘does someone know how’ ▪ Functional Compatibility- Impersonal pronouns ▪ Empirical Support – 1/3  ‘does somebody know how’ ▪ Functional Compatibility - Impersonal pronouns ▪ Empirical Support – 0 ▪ Pattern still retained – there might be support for somebody later on in the iterative process  ‘does john know how’ ▪ Pattern discarded Functional Compatibility from a subset of LIWC 1 -Cognitive mechanical (e.g., if, whether, wondering, find) ‘I am thinking about getting X’ -Adverbs (e.g., how, somehow, where) -Impersonal pronouns (e.g., someone, anybody, whichever) ‘Someone tell me where can I find X’ 1 Linguistic Inquiry Word Count,LIWC,

 Over iterations, single-word substitutions, functional usage and empirical support conservatively expands S is  Infusing new patterns and seed words  Stopping conditions

 doesanyoneknowhow  anyoneknowhowto  idontknowwhat  knowwhereican  tellmehowto  idontknowhow  anyoneknowwherei  doesanyoneknowwhere  doesanyoneknowwhat  anybodyknowhowto  anyoneknowhowi  imnotsurewhat  doesanybodyknowhow  doesanyoneknowwhy  iwaswonderinghow  doesanyoneknowwhen  tellmewhatto  imnotsurehow  iwaswonderingwhat  noideahowto  someonetellmehow  havenocluewhat  doesanyoneknowif  idontknowif  knowifican  anyoneknowifi  imnotsureif  iwaswonderingif  ideawhatyouare  letmeknowhow  andidontknow  nowidontknow  butidontreally  waswonderingifsomeone  wouldliketosee  seewhatican  anyonehaveanyidea  wonderingifsomeonecould  waswonderinghowi  idonotwant

 Information Seeking patterns generated offline  Monetization Potential of a post calculated by  Finding its Information Seeking score : Extracting and comparing patterns in posts with extracted patterns +  Finding its Transactional Intent Score: Using the LIWC ‘Money’ dictionary ▪ 173 words and word forms indicative of transactions, e.g., trade, deal, buy, sell, worth, price etc.

 Using a training corpus of 8000 user posts  MySpace Computers, Electronics, Gadgets forum  Generated 309 unique new Information Seeking patterns  Test Set: Using 3 sets of 150 posts each from Facebook ‘to buy’ Marketplace  All these posts have Information Seeking and Transactional intents  81 % of these posts were identified as monetizable in nature using our algorithm  Validates usefulness of action patterns

Off-topic Noise Elimination from posts with Monetization Potential

 Identifying keywords in monetizable posts  Plethora of work in this space  Off-topic noise removal is our focus I NEED HELP WITH SONY VEGAS PRO 8!! Ugh and i have a video project due tomorrow for merrill lynch :(( all i need to do is simple: Extract several scenes from a clip, insert captions, transitions and thats it. really. omgg i cant figure out anything!! help!! and i got food poisoning from eggs. its not fun. Pleasssse, help? :(

 Topical hints  C1 - ['camcorder']  Keywords in post  C2 - ['electronics forum', 'hd', 'camcorder', 'somethin', 'ive', 'canon', 'little camera', 'canon hv20', 'cameras', 'offtopic']  Move strongly related keywords from C2 to C1  Relatedness determined using concepts of information gain  Counts from Web as a corpus  Makes for a domain independent solution

 C1 - ['camcorder']  C2 - ['electronics forum', 'hd', 'camcorder', 'somethin', 'ive', 'canon', 'little camera', 'canon hv20', 'cameras', 'offtopic']  Informative words  ['camcorder', 'canon hv20', 'little camera', 'hd', 'cameras', 'canon']

Ongoing Work

Ideally, we would like to deploy on SNSs and observe click throughs Approximating with subgoals 1. Effectiveness of using topical keywords instead of entire post content 2. Effectiveness of using user generated content on SNSs instead of profile (homepage) information

 Keywords from 60 picked monetizable user posts  45 MySpace Forums, 15 Facebook Marketplace split into 10 sets of 6 posts each  30 graduate students, each set of 6 posts evaluated by 3 randomly selected users

 Google AdSense ads for user post content vs. extracted topical keywords

 Choose relevant Ad Impressions  VW 6 disc CD changer  I need one thats compatible with a 2000 golf most are sold from years if anyone has one [or can get one] PLEASE let me know!

 Users picked ads relevant to the post  At least 50% inter-evaluator agreement  For the 60 posts based on content  Total of 144 ad impressions  17% of ads picked as relevant  For the topical keywords  Total of 162 ad impressions  40% of ads picked as relevant

 User’s profile information  Interests, hobbies, tv shows..  Non-demographic information  Submit a post  Looking to buy and why (induced noise)  Qsn asked: Select ads that generate interest, captured attention

 Using profile ads  Total of 56 ad impressions  7% of ads generated interest  Using user submitted posts (entire content, already monetizable)  Total of 56 ad impressions  43% of ads generated interest  Using topical keywords from submitted posts  Total of 59 ad impressions  59% of ads generated interest

 User studies small and results preliminary, but clearly suggest  Monetization potential in user activity  Improvement for Ad programs in terms of relevant impressions  Evaluations based on forum, marketplace  Verbose content  May not work as well for micro-blog like content, status updates etc.

 A world between relevant impressions and clickthroughs  Objectionable content, vocabulary impedance, Ad placement, network behavior  Our works fits in a pipeline of other community efforts  No profile information taken into account  Cannot custom send information to Google AdSense

 Social Media Content Kno.e.sis Social Media Content Kno.e.sis  Google/Bing: Meena Nagarajan    Google/Bing: Amit Sheth    Sponsors: NSF (Semantic Discovery - SemDis), IBM UIMA Innovation Award 2007: "UIMA-based Infrastructure for Summarizing Casual, Unstructured Text”, Microsoft's Beyond Search - Semantic Computing and Internet Economics Award 2008: Chatter, Intent and Good Karma for Targeted Advertising in Social NetworksSemantic Discovery - SemDisChatter, Intent and Good Karma for Targeted Advertising in Social Networks