Natural Language Processing and Generation Tools to Consider

Robot Writer

By Trevor Clarke, TRA Director

For at least half a decade a considerable portion of basic financial news reports put out by the major news wires have been written by robots.

At the same time on social media like Reddit, Twitter, Facebook and Tinder it's likely you've had a "conversation" with a bot at one point or another.

And, of course, you may have already spoken to Cortana, Siri, Alexa, or another "assistant" on one of your devices.

These are just a few of the increasing number of examples where natural language processing (NLP) and natural language generation (NLG) are being used. There are many more. Some of them are just basic automated scripts, and others are "intelligent" to some level - they learn and can amend their responses.

We expect the use of NLP and NLG to proliferate significantly as organisations look to how they can automate the processing of human and machine information (the NLP part) and turn it into output (the NLG part), whether this be as text or a conversation of the verbal or written type (perhaps even sign language).

The following is a list of some NLP and NLG providers that have caught our attention. It's very likely we've missed a few (or many) so we'll add them as we find them.

One of the key things to keep in mind is that there are open source frameworks or libraries you can use to build your own tools, or you can consume "as-a-service". The market for NLP and NLG is maturing rapidly and as such there are several ways to go about implementing them in your organisation.

Another point is that NLP and NLG are often best applied when you have a repeatable set of inputs and outputs. The tools you use will have to be trained over time, so these technologies and services do not often suit ad hoc requirements.

Arria: Offers an Articulator Pro, for complex applications, and Articulator Lite for more basic needs. Also offers services. The company is also building out its own NLG-based SaaS applications such as an accounting reporting tool and has tied up with IBM Watson.

Automated Insights: Has many reference customers, including Associated Press who it has worked with on automating financial reporting for many years. Also offers an Excel plug in for trialling basic NLG with charts and diagrams. Provides managed services as well.

Yseop: Offers Compose which can be consumed on-premises, from the cloud or as a service. Also offers Savvy, a plug in for Excel and other analytics platforms like Qlik.

Narrative Science: Platform is called Quill and is complemented by managed services. Has many reference customers across industries.

AX Semantics: Plans starting at 199 Euro per month, AX Semantics says it offers eCommerce, journalistic and data reporting (e.g. BI or financial reporting) NLG services.

NLTK: An open source natural language toolkit written in Python.

GATE Cloud: Positions itself as a Text Analytics as a Service provider with a free tier of up to 1000 API calls per day.

Simplenlg: An open source and more basic (than some other services above) Java API for NLG written by the founder of Arria.

Core NLP: Stanford University's NLG offering, which is open source for non-commercial usage. Reputed to be suited to large scale requirements due to its speed performance.

OpenNLP: An Apache Foundation project with plenty of developer support. Includes integrations with other Apache offerings like Solr.

Microsoft: As part of its cognitive services, Microsoft offers a range of different NLP and NLG based APIs.

IBM Watson: One of the first to offer an "as-a-service" model for NLG and other cognitive services with many examples of customer case studies.

AWS: Lex and Polly are the two main AWS offerings and are based on the same technology used by Alexa.

Google: Similar offering to the other major cloud providers and also has a free tier for trying the service out.

SpaCy: A new start-up offering NLG services in a SaaS model. Based on Python.

Salesforce: As with many other major software providers Salesforce has built NLP/NLG into its AI-related services and go to market.