Interview with Ricardo Baeza-Yates, CTO at NTENT

What is NTENT’s current mission?

Our vision is to be the standard bearer for searching and surfacing credible information.

What is your background and why is it a good match for NTENT?

My background is mainly in data science by its current definition. In the 1990s I began doing research on web search and continued with data mining, natural language processing and applications for machine learning over the last two decades. During that time, I co-authored a book with Berthier Ribeiro-Neto, the head of engineering for Google Brazil, entitled Modern Information Retrieval: The concepts and technology behind search [1], which won the Book of the Year award from the Association for Information Science and Technology (ASIS&T) in 2012, and remains the most cited book on information retrieval. I have also published hundreds of papers on these topics that include some of the first papers that used machine learning in search, and a best paper award winner for an innovative distributed search architecture. 

Walk us through NTENT’s products. Who do they serve and what problems do they solve?

Users have become complacent when it comes to consuming biased information. By partnering with Credibility, NTENT provides businesses with a search and evaluation tool that helps content providers, affinity groups, and telcos make it easier to protect their brand, drive engagement, grow revenues, and surface reliable information. Companies are able to deploy the technology easily, maintain brand alignment, harness built-in revenue streams, and elevate the trust level of the content people see, beyond human bias. 

What are the biggest challenges (in business and technology) that you are currently facing at NTENT?

    In business and technology, we consistently face two predominant challenges.  First, we face the classic challenge of data and user scalability. We need to be able to handle more data as the number of users rises. To combat that challenge, we use a balanced combination of good algorithms and distributed systems. The second obstacle we face is search quality, partly due to the high amount of data. This calls for contextualized ranking based on machine learning that can handle the volume, data, and user diversity.

How can enterprises leverage (web) search to provide a better customer experience?

    In many ways, but the two main avenues are via website search and vertical search. A good website search allows companies to learn what customers are looking for, but not finding, on your site. These include new products, alternative topics, problems that need answers, etc. Vertical search allows your customers to search in specific (private) data collections to leverage your content via the Web or an app. 

Why should businesses care about incorporating search into their strategy? What are some ways to go about it?

    Search can reveal both the strengths of a business and what it’s missing, the latter of which represents potential ways to develop your business. It also improves the customer experience, through features like query intention prediction and contextual recommendations (task, person, product, content, etc.) These could even include external recommendations since people return to a website not only because of good service but because you are also a good referrer. In fact, search is the ultimate referrer.

How has the search industry changed since it began and where do you see it going in the future?

    The search industry started in the ‘60s with libraries, continued later with the first business applications, and culminated with the explosion of the Web at the end of the ‘90s. Today search is everywhere. We are only at the beginning when it comes to smart search assistants, but they will become ubiquitous in the future. To prepare for this, we need to integrate the Web and the world of IoT, improve query intention prediction and query assistance, provide better contextualization (mobile search), data integration (starting with personal data), and implicit search (assistant searches for you). We forecasted many of these topics almost ten years ago [2].

What type of role does AI play in search? How does NTENT use AI in its products?

    Machine learning, the main successful subfield of AI, is used everywhere in search. Primarily it’s used in ranking, with techniques found based in deep learning and learning-to-rank. It is also used in intention prediction, entity detection, spelling correction, query auto-completion, and the recommendation of similar items, among other functionalities. 

Looking ahead, what are the most interesting business use cases for search over the next five years?

    There are many interesting business cases. We are currently working on two of them. 

The first is fair search, meaning we are working to ensure that search results are gender and race-neutral, diverse and inclusive, and politically balanced, among other possible biases. This is crucial not only in news, but in all topics where ethics is relevant and biased results can harm people.

The second is conversational search, where we are able to handle conversations that are mainly search-driven, made up of different contexts such as e-commerce and home automation. The challenge here is that the context and ambiguity that arises when multiple people engage in multiple conversations should be carefully handled.

[1] R. Baeza-Yates, B. Ribeiro-Neto. Modern Information Retrieval: The concepts and technology behind search. Second edition. Addison-Wesley, UK, 2011. 

[2] R. Baeza-Yates, A. Broder and Y. Maarek, The new frontier of Web Search Technology: Seven challenges. In Search and Computing, Trends and Development, S. Ceri and M. Brambilla (eds). LNCS 6585, Springer-Verlag, Berlin, 2011.

About The Author

Scroll to Top
Share via
Copy link
Powered by Social Snap