Skip to content
No description, website, or topics provided.
Branch: master
Clone or download
Latest commit 26f6b42 May 16, 2019
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
.gitignore adding png result Feb 26, 2019
README.md Update README.md May 16, 2019
journal_reccomendation_1.ipynb styling modification Feb 26, 2019
journal_reccomendation_2.ipynb styling modification Feb 26, 2019

README.md

Scientific Journal Recommendation:

Installation:

  1. Download NLTK word corpus
  2. Install wordcloud
  3. Install tensorflow

Prerequisites:

  1. Datasets: link - https://archive.ics.uci.edu/ml/datasets/NIPS+Conference+Papers+1987-2015
  2. Tensorflow installation in anaconda: conda install -c conda-forge tensorflow
  3. Wordcloud installation: conda install -c conda-forge wordcloud
  4. nltk download: conda install -c anaconda nltk

Tensorboard Command and localhost :

  1. tensorboard --logdir=file_path --port=5000
  2. http://localhost:5000

Description:

This a unsupervised reccomendation system of the Scientific Journals. Firstly we have imported the libraries and dataset processed the raw text and saved the process dataset into a new csv file.All this operation is done in journal_reccomendataion_1.ipynb

On the second file, operations like removing missing values takes place.Then we do Exploratory data analysis of the data set like word cloud clustering and 2D t-SNE and also used pyLDAvis to visualize the cluster and most occuring word and to predict the suitable number of cluster.

We have used Count vectorizer to create a word vector which is fed into Latent Dirichlet Allocation,which gives us likelihood of the document to occur in the given topic cluster.Then we try to find the root mean square error to predict between the given document with all other document as a result we get the R.M.S Error for every comparation. We sort the given error in ascending order and reccomend 10 least errorneous title as a reference for the research paper along with error value.

Result:

crop

Contacts:

(For any queries mail_to: sharmaanix@gmail.com)

You can’t perform that action at this time.