We have all heard the sentence “This call may be recorded for quality and training purposes” when you call the call-centre of a company. Although some calls are indeed used for training purposes, more often they are used to improve natural language processing algorithms. The data from these calls help to create statistical models of phrases and words to improve the automated services to customers calling. This Natural Language Processing is gaining enormous traction and has massive potential for organisations.
In short, natural language processing (NLP) gives machines the ability to read and understand as well as derive meaning from the languages humans speak, and it is part of Artificial Intelligence. The challenge with NLP lies in the fact that computers normally requires humans to speak the programming language, which has to be unambiguous and highly structured, while natural language is anything but unambiguous. The linguistic structure changes with many complex variables such as regional dialects, context or slang language.
NLP is at the heart of modern software that processes and understands human language, leveraging the vast amount of language data on the web and in social media. There are ample examples of NLP but one of the most well known is probably Watson, who was able to win the TV show Jeopardy by beating two of Jeopardy’s greatest champions. Nowadays, IBM has opened up Watson’s cognitive computing intelligence to spur innovation and fuel new applications.
In a post on Chronicle, Geoffrey Pullum – a professor of general linguistics at the University of Edinburgh, describes three pre-requisites for computers to master to be truly able to understand natural languages. According to him these are “enough syntax to uniquely identify the sentence; second, enough semantics to extract its literal meaning; and third, enough pragmatics to infer the intent behind the utterance, and thus discerning what should be done or assumed given that it was uttered.” Although syntax and semantics are well underway, as there are ample examples of how this can be done, the hard part is the pragmatics as often a lot more is told by what is not said in a sentence or conversation than what is said.
However, despite the difficulties in mastering pragmatics, a lot has been done in the past decade and natural language processing has come a long way. In many areas natural language processing has find its way and we now talk to our phone to do things (Siri), talk to our browser to find stuff (Google) or talk to our car to change radio channels or call someone (Ford recently updated it cars with understanding 10.000 words or phrases instead of the initial 100 words). The Defense Advanced Research Projects Agency (DARPA) is currently testing language translation that can keep up with speech and is capable of translating English to Arabic and vice versa. Being able to understand spoken language is just one part of the deal, as being able to dissect written language is also an important aspect. This could lead to applications such as predicting the stock market based on how cheerful people are on Twitter. All require at least an understanding of the syntax and the semantics.
Talking to computers still requires programming languages but the ultimate goal of NLP is “to do away with computer programming languages altogether. Instead of specialized languages such as Java or Ruby or C, there would only be “human.”” The better a computer is capable of understanding natural language, the more applications there will be, but already natural language processing offers a lot of possibilities to help grow your business, save money and stay ahead of your peers.
Of course the best example would be the integration of sophisticated natural language processing algorithms in your call centre, which could result in a reduction of call centre staff, saving you a lot of money. Already many calls in call-centres around the world are automatically routed using voice recognition software. Or what about doctors who use speech software to record and transcribe information related to patients’ visits or medicines. The ability to understand difficult medical terminology has become a lot better in recent years.
Microsoft’s Managing Director Eric Horvitz uses a digital assistant at the reception outside his office. The digital avatar is a full-fledge assistant that is capable of welcoming people, checking appointments and calendars and tracking computer or cell-phone usages to being able to estimate how long a call will take before Eric is available again. The researchers working in this project even plan the digital assistant to be able to chitchat with waiting colleagues.
Natural Language Processing can also help to better know what your competitors are doing. By having algorithms listen to customers calling your call-centre and detecting how often names of competitors are mentioned, you can be alerted if a rival has, for example, lowered their prices. GoDaddy, an Internet Services Provider, found out this way that customers did not understand how to use GoDaddy on the iPhone and was able to quickly create instruction scripts for their agents.
Of course, when we use voice-recognition software while NLP is still not yet perfected it could result in mistakes that will be made. Until we have reached a level that we can naturally speak to computers like in Star Trek, which Google predicts will be reached within five years, human supervision will still be required. But despite the mistakes that occur, Natural Language Processing could already significantly impact your financial results positively. With the vast amounts of language data that is fed into the algorithms these days, Natural Language Processing is improving rapidly and there will be interesting applications ahead of us. If you are looking to integrate natural language processing within your organisation, here is a list of 25 API’s, collated by Mashape, that bridge the human-computer interaction.