3. Phases
Extract tweets and Extract Semantics Integrate and Design
user meta • Consume Alchemy web user Interface
• Retrieve 100 most recent service to receive • Integrate all modules
tweets semantic metadata of • Desing user initerface
• Retrieve list of friends tweets and the links in
tweets • Perform testing
and friends
• Extract links from each • Use cosine algorithm to
tweet determine the similarity
• Implement google
similarity and computer
aggregate similarity
4. API and Web Services
Twittter Rest API
Get tweets
Get friends and followers of a user
Alchemy REST API:
Get semantic meta-data for the tweets and contents of links in
tweets
Readablility API
Clean up messy HTML before sending requests to Alchemy
7. Workflow
1. Get tweets and user metadata
2. Extracts urls and get url contents
Two twitter
usernames
3. Clean up html contents
User Timilar
4. Get Semantic Metadata
1. Tag Cloud
2. Similarity Chart
3. Similarity Score 5. Compute Keyword and Semantic
Similarity using Cosine Algorithm
6. Compute Google Similarity for
each pair of terms
13. Demo
Web Application
Input – two twitter usernames
Output –
Tag cloud
Keyword Similarity
Semantic Similarity based on Alchemy API
Aggregated Semantic Similarity using NGD
14. Accomplishments
Finding similarity between any two entities
instance matching
semantic associations through
schemas/vocabularies/ontologies/classifications
Metadata and ontologies for connecting disparate data
Latent Semantics
Similarity Algorithms
Jaccard’s Coefficient
Cosine Similarity
Google Similarity