Personalization is as important for busy researchers looking for right reads as it is for shopping online

Personalization has been the game-changer for e-commerce platforms, transforming how people discover and buy products. Regular users can not only search for what they want, but also usually receive recommendations that simplify the process and reduce effort. You and I no longer need to browse endless options before making a choice, but unfortunately, researchers still do. Researchers must keep up with the latest developments and focus areas in their field to produce impactful results and advance science, which makes it critical for them to find and read the right papers. However, sifting through millions of published research articles on different databases, library websites, journals, Google Scholar and so on is a tedious, time- and effort-intensive process. While e-commerce uses personalized recommendation tools to save time and improve overall experience, the use of such tools to aid research discovery has been lacking! Moreover, researchers cannot simply browse papers as is possible with products on e-commerce sites. Trying to scan and understand a research paper often requires some mental gymnastics. According to a 2019 study by Elsevier, researchers spend about nine hours a week on research discovery and reading, with 50 per cent of this spent finding papers to read. Even then, only about half the research papers read are relevant, which means 2-3 hours of a researcher’s time is wasted every week. To add to researchers’ woes, this ocean of published research expands by roughly two million papers every year. Existing solutions for researchers Rapid technological advances are always knocking at the researcher’s door, but solutions available to help them find papers are far from optimized. There is much room for improvement. Imagine if you were buying books on Amazon or watching a movie on Netflix. What should happen after you finish reading a book or watching a movie? Would you want to start the search for a new book or movie from scratch, or should the platform know your preferences and suggest the next set of books or movies for you? Most of us would say the latter. However, if systems work only on search mechanisms, it raises two problems: (i) Problem of repetitive search, where users have to search for the same thing again in multiple sessions to proceed with their journey. (ii) Problem of stateless search, which means that user sessions remain the same unless a user changes the search query or filter results, or new results rank higher. Ideally, no two sessions should be the same for a given user because s/he will be at two different points in his/her research or reading journey. In the above example, an intelligent product would save your personal details and use it to suggest books or movies you are likely to find interesting. A similar intelligent product for researchers should be able to find and suggest relevant papers to read without users being required to do a search every time they use the product. The smart product should also be able to offer intelligent results that will help researchers improve and make progress on their own work. For instance, researchers would be interested in reading previously published work that has been cited in a particular paper, research that builds on their research subject, or even other research papers adjacent to the topic. A good recommender tool should be intelligent enough to learn from a user’s reading history and actions and suggest the next best, most relevant papers to read. Here, recommendations are different from search experiences; while users can use search functions to find papers, the system should have an additional option for users to discover newer content with updated recommendations. Personalization is the solution for researchers There is a visible gap in terms of products that can find and recommend the right scholarly papers for researchers to read. Technologies are emerging that can continually learn and improve recommendations based on user actions, which means the possibilities are endless. While researchers may still want to search for topics or papers, recommendations save time and eliminate the need for repeated searches on the same topics. And since such machines have more computational power than human brains, they can even recommend related topics that researchers might not be tracking but should be, which boosts one’s productivity. Building personalization into research products There are three broad ways to build more personalized products for researchers. 1. First, the product must have the intelligence to access a data repository and identify research papers that are best matched to user interests. 2. Next, it should introduce additional algorithms to understand user data and feedback and use this to generate personalized recommendations. 3. Finally, the product should be able to recommend related topics and journals that are relevant to users but are not currently being followed by the user. For optimized reading based on user feedback (step 2), products not only need quantitative data, they must also engage with users for more qualitative feedback on the relevance of recommendations. Potential gaps can then be addressed to improve the recommender systems. Listening to researchers and deploying corresponding improvements ensures the product can lead to improvements in recommendations over time. The impact of these efforts can be measured using metrics like the distribution of positive and negative feedback on recommendations within the product. Long way to getting personalization right Much like e-commerce, personalization is key to helping researchers navigate the ocean of research published in their field. However, there is a long, challenging road ahead before this becomes a regular feature. While many existing products claim to be personalized, they fail to manage basic functions such as maintaining an updated database, eliminating predatory titles, removing ambiguities around journal/author names, and so on. Products must focus fundamentally on researchers, while coming up with sustainable solutions that can not only personalize but improve the research discovery and reading experience for academics. While this may be daunting, it’s how technology can contribute to rapid advances in the research world. After all, the well-known facts of today are the result of yesterday’s research. The author is CTO, CACTUS. Views expressed are personal. Read all the Latest News, Trending News, Cricket News, Bollywood News, India News and Entertainment News here. Follow us on Facebook, Twitter and Instagram.