ResearchPaperRecommendation - FindIt

FindIt is a software project designed to recommend research papers and articles based upon a given query.The system currently uses tfidf algorithm to extract the best score papers from the database.

Prerequisites :

mongodb

Creating the database :

1.Go to the directory where mongoDB is installed and start the mongoDB service to open the mongo shell

$cd /path/to/mongodb
$sudo systemctl start mongod
$mongo

2.Create a new database "recom"

$use recom

3.Quit the mongo shell and import the datasets into your database.The following commands create three collections-'papers1','papers2' and 'papers3'

$quit()
$mongoimport --type csv -d recom -c papers1 --headerline --drop ..path/to/dataset/1_sorted.csv
$mongoimport --type csv -d recom -c papers2 --headerline --drop ..path/to/dataset/nlp3.csv
$mongoimport --type csv -d recom -c papers3 --headerline --drop ..path/to/dataset/nlp2.csv

4.To close the mongodb server,use the following command

$sudo systemctl stop mongod

Setting Up virtual Python3 environment

1.Follow the commands to create a environemnt "env" and activate it

$mkdir env
$python3 -m venv env
$source ./env/bin/activate
$pip install -r requirements.txt

2.To deactivate,use

$deactivate

Searching papers on the basis of string matching

Make sure the mongo server is active.Also activate the virtual environment and run the following command,

$mongo
$source ./env/bin/activate
$python web_nlp.py

Follow the local url displayed.A FindIt website will be hosted.Create an account if you do not have one.User need to sign in to go to the search query page.Option is also provided to make changes in profile settings including uploading a profile picture and changing profession and workplace.

A list of papers is displayed.You can add more papers into your database and make it extensive. A Url search option is provided which enables to open the desired paper by typing its Url in the search bar.

Future Work

1.Presently the system does not take into account the previous searches of the current user and those of other users.The algorithm will be further updated to take this into account

2.Optimisation of existing algorithm will be done to improve upon the searching speed.

3.The system only displays the paper topics and urls and provide the url search option using a search engine.A download paper option will be provided to enable the user to download in case required.However lack of storage space is the main problem hindering this functionality

4.The abstract of the papers is taken into account while computing tfidf score.The abstract and title data are computed manually and stored in the csv file.Code for performing this task will be updated.

5.User should be allowed to upload paper if he desires. No uploading option has been provided yet.The user can contact me in case he wants to add some necessary paper into the database but cannot upload on its own.So this feature will be added in future versions

6.More features will be added to make the system user-friendly