Annotation_Interface

About

• Python based GUI(Graphical User Interface) for annotation of documents.
• It annotates documents on a sentence level, assigning each sentence with the tags.
• Can be used for easing annotation of data for certain NLP(Natural Language Processing) tasks, like novelty detection.

Working

• After the environment is set up, run the integration.py python file
• The annotation interface opens, and prompts to select a text file.
• The senteces are tokenized using the nltk tokenizer. Each sentence can be assigned with various tags.
• The four criterion of classification include: -
  • Sentiment Layer
  • Section Aspect
  • Subjective Polarity
  • Is major comment
• The Section Aspect Layer is multi choice, wheras all the other layers accept a single option.
• To chose other, type the information in the box and tick(select) the other button.
• Click on next to proceed anotating to the next sentence.
• After all sentences are done, a new file with the <file_name>_anotated.txt will be stored in the same location as that of the parent file.

Window description

• Select the file from the file explorer(opens on clicking browse option) and click okay.
• Sample preview of the window.

Requirements

• Python 3
• nltk package(natural language toolkit)
  • pip/conda install nltk
• Download punkt tokenizer module of nltk
  • Type 'nltk.download('punkt')' in the python terminal after importing nltk
• wxpython(cross platform gui toolkit for python)
  • Link for downloading OR
  • 'conda install -c anaconda wxpython' in the terminal for conda users