Domain-Specific-Keyword-Extraction-using-Spacy

• Named entity recognition (NER) , also known as entity chunking/extraction , is a popular technique used in information extraction to identify and segment the named entities and classify or categorize them under various predefined classes.
• In any text document, there are particular terms that represent specific entities that are more informative and have a unique context. These entities are known as named entities , which more specifically refer to terms that represent real-world objects like people, places, organizations, and so on, which are often denoted by proper names. A naive approach could be to find these by looking at the noun phrases in text documents.

Goal

• Customized NER model is created using NER to find domain specific Keyword

Background

• NER plays very important role in information extraction.
• When we have large corpus containing different keywords related to different domain,NER comes in action in those scenarios.
• NER can be beneficial in news headlines.

Data Preparation

• Data preparation is the main challenging task in NER.
• In this project , I have created a corpus of 65 records.
• data.txt contains 65 sentences.
• labels.csv contains entities and labels

Model Training

• Customized model is created using spacy to find domain specific keyword.
• Here the domain specific keywords are food,cloth.
• The user is asked to enter the model name and the model is saved with the same name in local directory.

Testing

• The user is asked to enter the text.
• The user will get domain specific keyword:food,cloth based on the text

SetUp

• pip install spacy

Steps to follow to execute the project

• clone the repository into your local system
• Run the command: python Nermodel.py
• Note You can change the number of iterations based on your requirement, I have kept 20.
• After the completion of model training, you need to enter the model name and the model is downloaded into your local directory where the code is stored.
• Finally , you need to enter the text and you will get the keyword based on your text.

Future Work

• I have used only 65 sentences to train the model, you can use more corpus and train the model.
• You can use as many lables as you want,you can add more lables
• You can use NER in news headlines and you can extract domain specific keyword from the news.