A custom developed collection of text and natural language processing components.

We have designed solutions components that can be quickly adapted to create custom products to address specific business needs. These are developed using a combination of IBM® SPSS® modeler and other custom components.

Sentiment Mining

Sentiment mining enables an organization to intelligently sift through customer communication (emails, web postings, discussion forums etc.) to quantitatively characterize the subjective opinions or emotions of the customers towards a particular product, topic, or the company in general.

By developing a custom library of natural language processing components developed using SPSS®, customer communication can be processed to produce the underlying structures that feed the sentiment scoring model. The natural language preprocessing components include:

  • Parsing and tokenization
  • Sentence boundary detection
  • Part of speech tagging
  • Phrase chunking
  • Sentiment tagging

Once the text has been preprocessed using the components listed above, it can be scored for sentiment either using our general-purpose sentiment scoring model, or a custom scoring model specific to the application domain.

Topic Mining

Topic mining enables an organization to automatically sift through customer communication (emails, web postings, discussion forums etc.) to summarize the common topics or threads that are been talked about to capture what is been talked about. For example, it can enable a company to sift through costumer postings to summarize the top complaints that have been discussed during the last week.

Using our custom library of natural language processing components developed using SPSS®; the customer communication can be processed to produce the underlying structures that feed the topic extraction model. The natural language preprocessing components include:

  • Parsing and tokenization
  • Sentence boundary detection
  • Part of speech tagging
  • Phrase chunking

Once the text has been preprocessed using the components listed above, a custom topic detection algorithm is applied to extract and characterize the most relevant topics in the discussion.

Both the sentiment mining and topic-mining products can work with multiple input formats, ranging from flat files, XML, or any ODBC compliant datastore.

To find out more please call us today. click here