This is due in large part to the rise of chatbots and intelligent assistants in call centers, help desks, kiosks, and other customer support applications, but these are hardly the only ways to apply NLP. Back-office functions ranging from software development and data analytics to systems management and risk assessment become far more efficient and effective when humans can simply speak their desires rather than type them in or click their way through endless menus.
Getting to that point won’t happen overnight, however. While NLP has taken great strides recently in terms of accuracy and efficacy, it still has some way to go before it becomes a valued member of the team.
NLP is following the money
Still, over the past year, the enterprise has displayed an increased willingness to open its checkbook a little wider to fund various NLP projects. According to new research by NLP developer Jon Snow Labs and data analysis firm Gradient Flow, 60% of tech executives reported at least a 10% increase in NLP funding, with about a third reporting jumps of 30% or more. Health care, technology, education, and financial services were at the forefront of this curve, while applications like name identity recognition and document classification were among the primary use cases.
NLP’s appeal lies largely in its capability to digest large amounts of unstructured data, which has long been suspected of housing crucial pieces of information and hidden data patterns that can do wonders for business development, productivity, and competitiveness if leveraged properly. Service Express data science manager Jim Carson noted on Data Center Frontier recently that NLP essentially fills the gap between computer understanding and human understanding. This can lead to significant improvements to a wide range of enterprise processes, such as email management and contract analysis, as well as equipment logging and data center infrastructure monitoring.
NLP can also make significant contributions to the enterprise when combined with other forms of artificial intelligence like machine learning. CIO.com recently highlighted the work of the Computational Story Lab at the University of Vermont, whose work in sentiment analysis builds on the integration of NLP, ML, and other techniques to glean the emotional context of communications. The lab’s Hedonometer project currently evaluates 50,000 tweets per day to calculate a daily “happiness score.” While the system currently uses a rudimentary plus-minus scoring system to reach its conclusions, more refined algorithms may one day be able to create more complex analyses and target specific data to track things like brand popularity and consumer trends.
A new understanding of NLP
Meanwhile, IBM’s Watson remains as one of the leading conversational iterations of NLP, and the company has added a number of new capabilities since the platform became a Jeopardy champion 10 years ago. It, too, is working on extracting more complex meaning from leading document formats, like PDFs, as well as advancing the fields of multi-language communications and empowering subject-matter experts with data analysis and knowledge development. It also sports new customization features that simplify AI training processes.
All of these developments seem to be trending in one direction: the development of a fully conversational user interface that makes accessing vast computing power as easy as chatting with a coworker. We aren’t there yet, but before long expect to see a new member of the enterprise team, essentially the enterprise itself, conversing at meetings, responding to user requests, and maybe even sharing a joke at the water-cooler.
As a new employee, it will have a lot to learn, but it is already showing tremendous potential.
This article was originally published in VentureBeat and is reproduced with permission.