Topic Modeling

Discover topics running through large collections of unstructured texts.

Details »

Sentiment Analysis

Quantify the positive and negative emotions expressed in texts.

Details »

Natural Language
Pre-Processing

Clean noisy texts through stopword removal, parts-of-speech tagging and lemmatization.

Visualization

Visualize topics and sentiments in intuitive ways.

Automatically extract and visualize topics and sentiments
from millions of text documents


Scientific Papers

Accelerate your literature review and
automatically analyze the content of papers

Corporate Reports

Analyze offical reports published by large organizations, e.g., financial reports, corporate responsibility reports, or corporate governance reports

News

Track the development of topics in the news

Social Web

Listen to conversations on online social networks,
forums, blogs and other social media channels

E-Commerce

Mine the content and sentiment of
product descriptions and customer reviews
Live Demo: Online CustomerProduct Reviews

Any other type of documents

Analyze any text stored in
Excel, JSON or TXT files

Our Algorithms

We apply only well-documented and scientifically evaluated algorithms.

Topic Modeling

Latent Dirichlet Allocation (LDA)

The idea behind topic modeling is that words that co-occur together in similar contexts tend to have similar meanings. Hence, sets of highly co-occurring words (e.g., ball, pitch, goal) can be interpreted as topics (e.g., football) and used to cluster documents into thematic categories. LDA is a popular topic modeling algorithm that is able to discover topics running through a large collection of documents and to annotate individual documents with topic labels. As an unsupervised machine learning algorithm LDA is purely data-driven and inductively infers topics from given texts — neither necessitating any manual labeling of documents, nor the existence of predefined categories.

Learn more

Papers using LDA
Sentiment Analysis

SentiStrength

Sentiment analysis deals with the quantitative measurement of opinion, attitude and subjectivity in texts. The SentiStrength algorithm estimates the strength of positive and negative emotions expressed in short texts. SentiStrength follows a dictionary-based approach to sentiment analysis and, therefore, can operate in many different domains. Besides relying on a dictionary of words with human sentiment polarity and strength judgments, it also exploits other (non-lexical) information, such as, negation, booster words, idioms, emoticons or punctuation.

Learn more

Papers using SentiStrength

Our Team

From researchers for researchers

Stefan Debortoli

Research & Development,
Product Management


Dr Oliver Müller

Research & Development,
Product Management


Professor Jan vom Brocke

Research,
Business Development


Michael Gau

Frontend Development,
Server Administration


Contact Us

If you are interested in using MineMyText for commercial purposes, please contact us.

MineMyText.com
c/o Stefan Debortoli
Staudachweg 12
6800 Feldkirch, Austria

© 2015 MineMyText.com