https://github.com/deven96/nlp-tasks

Compilation of NLP tasks
https://github.com/deven96/nlp-tasks

Last synced: about 1 month ago
JSON representation

Compilation of NLP tasks

• Host: GitHub
• URL: https://github.com/deven96/nlp-tasks
• Owner: deven96
• Created: 2020-02-29T00:08:10.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2022-12-08T03:42:26.000Z (almost 2 years ago)
• Last Synced: 2024-10-02T07:55:03.415Z (about 2 months ago)
• Language: Python
• Size: 42 KB
• Stars: 0
• Watchers: 3
• Forks: 0
• Open Issues: 7
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# NLP Tasks

This is a compilation of natural language processing tasks such as sentence similarity, named entity recognition, e.t.c

## Installation

```bash

pip install -r requirements.txt

```

## Data Sources

### Sentence Similarity

Get the [data](https://www.kaggle.com/quora/question-pairs-dataset), extract and store in `data/sentsim/questions.csv`

- `sentsim/wnn.py` makes use of a wide neural network to provide a sigmoid value of sentence similarity. 


The architecture used is shown below

![Architecture of WNN](assets/widenn.png)

To make use of it, pass both sentences to the script (NOTE: current threshold is 0.15)

```bash

python sentsim/wnn.py "I am very happy" "I am ectstatic" 

First sentence:  I am very happy 

Second sentence:  I am ecstatic 

Similar? :  True 

```

- `sentsim/doc2vec.py` makes use of a word2vec extension for paragraph, document similarity known as [doc2vec](https://www.quora.com/What-is-doc2vec?share=1)

### Named Entity Recognition

Get the [data](https://gmb.let.rug.nl/releases/gmb-2.2.0.zip) and store it as a folder `gmb-2.2.0/` under `data/ner/`