Machine Learning and NLP Enthusiast

Team Members

Launch Site

M-Analyzer

What would you like to find out in documents? M-Analyzer provides several services in the field of text analysis, such as text similarity for plagiarism detection, keywords and entities extraction, and aspect-based sentiment analysis

V800 team 0d8fc292adcadbb40bc2cd35a5e4f149 1493050502

Description

M-Analyzer provides several services in the field of text analysis, such as text similarity for plagiarism detection, keywords and entities extraction, and aspect-based sentiment analysis

Instructions

In the main page (left navigation bar), you can see the services provided by M-Analyzer. You can choose any services you'd like to use just by clicking on it. Afterwards, the corresponding form will be shown at the wider area on the right.

Plagiarism Detection


  • You can provide two documents that will be compared by the system. You must provide exactly two documents so that the system can proceed to the analysis part
  • After you click the button Analyze, the system will check the documents similarity. There will be a loader which means that the analysis process is still running
  • When the analysis is completed, you will see the document similarity analysis which is represented by some metrics, such as Cosine Similarity, Levenshtein Distance, Jaccard, Jarowinkler, and Tanimoto
  • Each metric will show a value ranging from 0 to 1 in which 0 means that both documents are absolutely not similar, whereas 1 means that both documents are absolutely similar

Keywords and Entities Extraction


  • You can extract only important aspects of information from any documents by using this service.
  • To use this service, you need to provide the URL of document or just the document directly. Please make sure that you have put the URL or document in the correct field (URL field is in the above of document field).
  • After you click the button Analyze, the system will start the extraction process.
  • There are two result categories provided by this service, namely Keywords and Entities.
  • For the Keywords category, the result provided by the system includes the extracted keyword, relevance score, sentiment score, and emotions score. The emotions score is divided into several categories, such as sadness, joy, fear, disgust, and anger.
  • For the Entities category, the result provided by the system includes the entity name, type of entity, sentiment score, relevance score, disambiguation score, and total amount of that entity in the document. The disambiguation score is divided into two categories, namely subtype (ex. AdministrativeDivision, GovernmentalJurisdiction, Country) and name.

Sentiment Analysis


  • You can find the subjectivity score of certain aspect of information by using this service. This is an aspect-based sentiment analysis in which it can analyze the sentiment of any aspect in the document. For example, if you have a document talking about market, you might want to know the sentiment score of market stock. This market stock is one example of aspect whose subjectivity score can be determined by this service.
  • The input is similar with the previous service, except for the input of Aspects. You can provide any aspects in the document you'd like to know their sentiment score. The maximum total amount of aspects is 15 and each aspect must be separated by , (comma).
  • After you click the button Analyze, the system will start the analysis process.
  • The system will provide two result categories, the first one is showing the sentiment score and label of document and the second one is showing the sentiment score and label for each aspect. The label states the sentiment category, such as negative, neutral, or positive.

View the screencast:
https://www.youtube.com/watch?v=C-lRw3gvJ-k

Built With

I built this app using these awesome technologies:

  • Node.js
  • AngularJS
  • Bootstrap
  • Watson Developer Cloud - Natural Language Understanding

Because I used Node.js, here are the packages (node modules):

  • express
  • body-parser
  • cosine
  • path
  • wuzzy
  • watson-developer-cloud

Feedback / Ratings (12)

All Feedback