By Jen Doyon – Director, Ntrepid Language Engineering (Ntrepid Corporation)
I have been working in the field of computational linguistics for the past twenty years. Up until a few years ago, I was a program manager, consultant and evaluator of natural language processing (NLP) technology solutions. I worked with technologies such as information extraction (IE), information retrieval (IR), machine translation (MT) and translation memory (TM).
My favorite part of the job was working directly with analysts, translators and their managers in order to establish their NLP challenges and provide them with relevant solutions. This has provided me the unique experience of being able to establish relationships with the users, buyers and developers of numerous academic, commercial and government technologies. It was through these relationships that I have become all too familiar with the opinions and attitudes of these three groups of people toward NLP technology and each other. So five years ago, when my boss came to my colleague Dr. Linda Moreau and me offering us the chance to propose an NLP research and development (R&D) project, we welcomed the opportunity.
From the beginning, we knew we wanted to design and build a tool that was intuitive and customizable by our target users. We decided our tool should fill a technological gap in the NLP world. Strategically, I wanted this tool to be the first in a suite of capabilities with a common focus. After hours of brainstorming, we had a list of potential ideas focused on the handling of named entities. The first item on our list was a translation engine for structured data. This would become Ntrepid’s Virtus TranslatorTM product.
In the past few years, I have evolved from a program manager to a product manager. There have been many lessons learned along the way and continue today. The biggest challenge for me was learning the intricacies of marketing a new product. Luckily, I had the help of an extremely patient and talented group of individuals to guide me through the process.
Now that our product is commercially available, I am thrilled to have reached our original goal. Like other MT systems, VirtusTM does not produce human quality translations. It was designed and built as a tool to help translate mundane and repetitive data, such as that found in spreadsheets and databases. It is user friendly and provides the ability for end users to customize almost every aspect of the system. In other words, the end user has control of what translations are produced by the system. Where traditional MT engines translate unstructured data, Virtus specializes in translating structured data or named entities. You can check it out for free at: www.ntrepidcorp.com/Virtus.
NOTE: The views expressed here are those of the authors and do not necessarily represent or reflect the views of GALA.