Natural Language Processing (NLP) Dictionary Server: A web-based system to create, edit, and manage NLP dictionaries.
The jStart team worked with our clients to develop a tool designed to help non-technical line of business users better manage NLP dictionaries, using a simple web-based interface.
Part of the mission of jStart is to aid the adoption of emerging technologies by filling technology gaps with those technologies. In the case of Natural Language Processing, enagements with clients indicated that a simple way to manage NLP dictionaries could have real business value. Because of this, the team created the NLP Dictionary Server.
what is an NLP dictionary?
NLP Dictionary: A list of like terms or phrases that is used to label domain concepts, like names, as a building block of text analysis.
Developing the solution
The NLP Dictionary Server has, at its core, two main functions:
- to help create, edit, and manage sets of dictionaries for NLP applications
- provide a real-time service for using those NLP dictionaries created.
The tool focuses on a very specific need: the ability to manage NLP dictionaries. "Previously, to build a dictionary was an intensely manual process for our clients. They had find the terms they wanted in their dictionary, then they had to go through the laborious effort of cleansing the data--making sure the text was in lowercase, accommodating variations in spelling, removing duplicates," noted Lee Surprenant, who lead the effort to develop the tool. "Basically we're trying to help automate the expansion of our client's dictionary to improve recall for their NLP models."
Direct client feedback
One of the unique properties of every technology developed by jStart is the fact that they are driven by client needs and requirements. That makes the technology instantly relevant. "We initially worked with a few clients to make sure that the tooling would work for them. What we learned is that the focus of the tool should be narrow, and the interface clean, simple, and straight-forward," explained Lee. Another insight provided by clients: the need for a simple versioning system--that didn't require you to be a developer to use.
In use today, future plans
The tool's latest version (1.1) is in use today by our clients. As part of the jStart process (explained in our article on the development team), jStart is busy listening to our client feedback. "We're looking to improve the experience of dictionary builders by incorporating instant feedback from the targeted corpus of data," remarked Lee. "The way we see it, there's a lot of room for innovation."
Resources and Links
Briefing. This briefing includes detailed information on how you can get started with the tool, including actual scenarios for use, and a hands-on tutorial on leveraging the tool. Schedule a Briefing.