We've developed and employed an authorship analysis system based in mathematical and scientific formulations. Our software and computational text analysis can be used independently or in tandem with traditional analyses to run a series of practiced methods in less time and with more accurate and robust results.
Using multiple analyses methods, we are able to validate our results since each method will likely direct us to the same result using a different formula.
Our methods for Forensic text analysis deliver on:
- Time – our computer can read a thousand page document in seconds
- Accuracy – the computer counts with much greater accuracy than the human eye, eliminating fatigue related errors
- Scale – the computer can handle large-scale document searching, with the ability to read and analyze 200,000 pages or more of documents for a single case
|
The methods used and our approach to text analysis should pass the Daubert standard, ensuring that our expert testimony and forensic analysis of your documents will not be subject to questions of validity. We can provide expert testimony from an accredited source with statistically based results of your analysis.
|
Our process is customized for each case that we work on, depending on the available text and the purpose for the analysis.
Broadly, our process follows a series of steps that can be best described as follows:
- Receive the document in question and any supporting documents to form a baseline
- Clean up the document and extract meaningful events or features
- Compare the documents events with other events extracted from the baseline document(s)
- Report the results
|
Upon review of your case specifics and documentation, we will develop a specific process for your needs. We will walk through that process and the various methods that we will employ and we will deliver results.
|
Our process and computational analysis program have been proven to work in all aspects of text analysis:
- Authorship Attribution
- Author Characteristic Profiling
- Plagiarism
- Large-Scale Document Searching
|
/Events
PAN 2012 Lab – Uncovering Plagiarism,
Authorship, and Social Software Misuse
September, 2012
During the PAN 2012 Lab –a competition to test various programs on the 1.5 million Enron emails released to the public to determine authorship correctly identified the author of 70% of the emails tested.
|
