Introducing RAx, a personalized assistant for researchers

(26 March 2018) Conducting a literature review is often the first set of skills required in postgraduate research. It is painstaking work to comb through the library’s physical collections and online resources. But this is changing. Researchers can now make connections with the literature, ideas and their peers using smart assistants. One such is RAx, an assistant developed by Rygbee Inc., (USA). ACCESS spoke to Sourish Dasgupta, Co-founder and CEO about RAx and began by asking,

Did you develop RAx for your own use?

Yes and no. RAx was developed by researchers for the research community at the Dhirubhai Ambani Institute of Information and Communication Technology, DA-IICT, India, where I am teaching. The development guidelines are based on my experience of more than 12 years of academic research, first as a Ph.D. student and then as a research professor mentoring many grad research students.  The target audience of RAx postgraduate students, including Ph.D. students, post-docs and research faculties. But anyone working on research projects will benefit from what RAx has to offer.

RAx is a virtual assistant like Siri and Alexa?

RAx is a personalized AI assistant that helps researchers to become more efficient at work. RAx can adapt to researchers’ changing needs throughout a research projects’ life-cycle. It does this in three ways. First, it helps researchers to discover and review resources based on their research topics. Second, it helps them to easily connect their newly developed insights with their previous reviews and current work-in-progress. Last, it seamlessly disseminates their findings for discussion with the project team members.

Why not use Google Scholar instead?

Google Scholar works well when the information needs of a researcher is a “theme” which is  expressed as keywords. However, in real life researchers spend a lot of their time reviewing the literature and going where it takes them. That is where understanding, reviewing, and connecting with previous insights and current work-in-progress occurs. That would include providing the right set of resources, often going beyond research papers (like research blogs and magazines, discipline-specific news sources, online lectures and tutorials, recordings of conference/workshop talks, and wikipedia articles) that address the needs of the research project. This is much more than a “theme” query, something that  a scholarly search engine is not designed to do.

How does the AI achieve this?

The AI engine behind RAx broadly works on four fronts. First, it understands how the theme of the content is evolving over time as the researcher reviews the literature and works on his or her project.  Next, it connects the content with other semantically related content by linking current work with past work. Further, it formulates search queries accordingly and does a semantic matchmaking exercise using a mix of several proprietary algorithms against some hundreds of millions of research-related resources. And finally, it reformulates the query by observing how the researcher is reacting to the suggestions given by RAx and how these suggestions are influencing further research work inside RAx.

Where does RAx get its content from?

There are several different resources that RAx currently indexes: research papers, research blogs, online lectures, conference proceedings, wikipedia articles, and discipline-specific news. We also have more than 160 million research paper abstracts. We are currently negotiating to partner with some of the top publication houses to index their full-text articles. Technically, any collection can be plugged into RAx.

How many users does RAx have?

In the first four months RAx has gathered more than 700 active researchers, mostly Ph.D. students from the United States and India. In total we have subscribers from many universities including MIT, Harvard, RMIT in Australia and the IITs India. RAx is offered as individual or campus-wide subscriptions.

How is support and development managed?

There is a feedback feature for the user which is accessible at all times. An email gets automatically triggered to our team and we take every measure to respond within 24 hours. For the development of RAx we are currently building research relationships with  RMIT-Australia and with an NSF funded Big Data Learning consortium at Carnegie Mellon University. Research is the soul of the development of RAx and is going to be the key driving force that will give us an advantage over others.

To know more about RAx visit the RAx website, https://raxter.io.

Request a free trial by writing to yauchan@igroupnet.com.