Skip to content

Latest commit

 

History

History
35 lines (22 loc) · 1.32 KB

README.md

File metadata and controls

35 lines (22 loc) · 1.32 KB

Makri - Malayalam Knowledge Ripper

A highly efficient data scraper with over 80,000 lines of malayalam data, which are POS tagged using a SVM algorithm for the creation of database of POS tagged lines for the use in Malayalam research and for Scholars.

Feel free to use the POS tagged data in the above files in anyway which doesn't violate copyright.

Technical Details.

The scrapper is build using Scrapy-Python. The makri-links.py can be used to collect links of malayalam articles and makri-sentences.py can be used to get malayalam text from websites and write into files which are divided by category.

The classification algorithm is made using nltk module in python\

A project by Adarsh S and Jithin James under the supervision of ICFOSS under the supervision of Dr. Rajeev RR

Web App

A Django based web app is built to distribute the service. The development server can be accessed via python2 manage.py runserver

The web app has an input text option for live data tagging, an upload file option to tag text files.

The sevice also has a web end-point to use in other applications.

curl -G -v "http://127.0.0.1:8000/" --data-urlencode "q=input" will return the tagged data for the given input.

Team Members

@isht3, @jjmachan -- creators of the project @abinmn -- deployable web app creator