Data-Tag

Developed By:

Akshay Pratap Singh(2011ecs01)

Rishabh Shukla(2011ecs13)

Sunny Kumar(2011ecs43)

Guided By:

Dr. Sunanda Gupta

Data-Tag is a Natural Language Processing based Text Classification system, which utilizes Open Data from Wikipedia along with Word-Sense Disambiguation algorithm to find semantic similarity and outputs tags accordingly.

What problem does it solve?

Provides an alternative for keyword based classifications
Tags data according to their semantic meaning
Makes text classifiers more artificially intelligent

What's new in Data-Tag?

Semantic Nature
Uses Wikipedia Data as gloss for finding semantic similarity
Can tag textual as well as data from web pages
extends capability of WSD to multiple classes
No Database requirements - using existing data from internet

User Interface/Demo

Technologies/Frameworks Used

Python - Flask
JavaScript - AngularJS
Natural Language Toolkit
Wikipedia API
Redis - noSQL Database
Pattern Framework
grrequests
urllib3

Sequence Flow

Send data from Angular Controller to Flask Back-end
Tokenize text
Apply Part-of-Speech Tagging
Extract Nouns
Get pages from Wikipedia based on extracted nouns
Apply Word Sense Disambiguation
Get matching scores
Return top 3 tags with highest scores

Sequence Diagram

Shortcomings

Slower processing on Slower networks
may provide inaccurate results for larger number of tags
No Memoization
Supports only "English" language

Scope/Future Work

Can be incorporated with a web crawler for semantically tagging web pages
Can be used with Machine learning techniques to provide more accurate results
May acts as a basis for NLP Query Formulation from web data

Conclusion

Systems like Data-Tag are the future of semantic text over internet. It aims in dropping age-old keyword based classification systems and welcomes the advent of more Artificially Intelligent systems. It further provides a basic structure to develop larger systems utilizing the similar concepts to classify data all over the internet and textual documents.