From problems to patents

From idea to feature

Doug: Natural language query processing and searching is a well-known problem in information retrieval technology. Natural language searching allows people to use normal, everyday language with a search engine rather than specific mathematical operations and rules to describe what they are looking for. LexisNexis was one of the first companies to give its customers natural language query capabilities in its products and has been constantly improving the algorithms used to provide our customers with better, more accurate results ever since.

Khanh: The problem we tackled was, how do we make search results in our Lexis Advance platform better so that the customer receives the top five documents relevant to their search? I want to make it clear that we didn’t start off with the idea of how we could create something that could be patented. We wanted to solve the problem of getting better results for customers.

Doug: A few years back, we found that we needed to work on natural language searching, and after some months, I had the idea for Query Token Densities (QTD). QTD is a search algorithm for evaluating and ranking the relevance of results from natural language queries. That means it’s a mathematical method for taking a search query, phrased in conversational words or phrases, then finding the best results for that query from a set of candidate documents. While there are numerous techniques for doing that, the novelty of QTD is that it assumes that when the query words are close together in a document, meaning a higher query term occurrence density, then that document is more relevant than documents where the query terms are spaced further apart. Khanh is a fantastic developer and was able to translate my idea into functioning test code very quickly. He also provided key insights on how to make the algorithm run faster and better on real data.

Khanh: We could tell when applying this algorithm that the customer got better results, and we thought it was unique enough that we filed a patent for it. The enjoyment I get from solving problems like this is hard to explain. You're actually just physically changing the orders of documents, but you're making customers happy because of it. It's a different type of happiness.

Making mental connections through play

Doug: For me, a really important part of creativity is putting things together just to see how they work with each other. Maybe you've got a very realistic portrait in part of the painting and a very abstract background in another and together they make the portrait really pop. Or maybe you layer a really strong melody over a barely there but complex percussive background that makes both stronger. Putting different simple things together frequently leads you into interesting and unexpected directions. The more you practice with this sort of approach, the more it's going to transfer over to how you think all the time. The hard part is keeping that “beginner’s mind” and looking for simple ideas that have profound possibilities.

Khanh: I'm not artistic in the sense that I can create art, but I like to use my mind. For instance, about four years ago I decided I wanted to learn French so I could speak more easily with my snowboarding friends in Quebec. I challenged myself, and now I can speak to my friends in French. It's about challenging my mind to keep it and my body young.

Filed Patent in Natural Language Query

US 20200341990A1 (Filing), System and methods for ranking electronic documents based on query token densities, D. T. Rosenoff & K. Vu

As well as the joint patent with Khanh Vu, Doug Rosenoff holds five additional patents in information retrieval.

It’s all about the team

I have been involved in quite a number of patents, but most notably is the Systems and methods for detecting tax refund fraud, which is a fraud detection tool used by individual state departments of revenue. This stemmed from a market planner bringing an industry-wide issue to my team as a product challenge for us to solve: how can we prevent tax refund fraud without disrupting the turnaround time of the states' processes? The idea for the solution came quickly, but working out the detailed “how” and fine-tuning took time. This was where teamwork, and having a knowledgeable group of varied subject matter experts, was of utmost importance. Once we had the solution, we filed for a patent.

I like to wake up every day knowing that our whole consolidated group is making a difference. Whether we are assisting on cases involving a bombing; looking for a terrorist, school shooter, or a missing person; or even if we are ensuring the correct person receives assistance and benefits, every single one of those situations is of extreme importance, not only to the individuals directly affected, but to society as a whole.

Making the connections

I think I've always had some sense of creativity, wanting to look for new ways to accomplish, create, or build things. Whether I was working with my brothers on the next best soapbox car or creating games to play with the neighborhood children, it was always about doing things more efficiently. At work, there is nothing I enjoy better than building new and creative solutions! I'll find myself reading or watching the news and hearing about different issues and problems that government entities may be facing, and it's just natural that I think, “why don't they use this service,” or “they could help prevent fraud by a different service.” To me, it's almost a point of frustration when you know that there are programs available. It's just a matter of reaching these agencies and bringing that knowledge to the forefront.

Patents in Fraud Detection

Systems and methods for detecting tax refund fraud

Systems and methods for improving computation efficiency in the detection of fraud indicators for loans with multiple applicants

Systems and methods for increasing efficiency in the detection of identity-based fraud indicators

Thriving in a culture of innovation

George: We were given the problem of creating a way to easily identify funding sources in research documents. This is vital information for government agencies and other funders to be able to track where the money goes and make decisions on funding future research.

Zubair: We had a process already, but it was manual, which was costly and prone to error, so the problem was very simple. Can we automate the process, do it better, fast, and at scale? These are the things that trigger creativity and imagination.

George: We needed an automated solution for the data extraction that could methodically search documents for the information. Through natural language processing, our invention reads and processes scientific text with the sole purpose of focusing on the parts of article that have to do with the funding information.

Be brave and patent!

Zubair: The freedom and space we have in RELX is unmatched. It’s the dream of researchers because we’re given a problem and asked to figure it out, and that is really an ideal atmosphere for people like us. We're always looking for cutting-edge algorithm techniques to solve problems. Usually our issues are very novel and haven’t been addressed by other research, but we don’t necessarily build an algorithm from scratch. We look at available alternatives, research, and collaborate to see how we can build on them to solve the business problem. It's about being brave enough to really go for the innovative solution.

George: It's a very long journey and process to patent a novel idea or system. It requires a lot of in-depth research and a deep understanding of the root causes of the problem. You must provide evidence that the system is applicable, implementable, and it produces the promised results with real data. And then, of course, others might come up with something similar, so you want to safeguard this idea and patent.

The whole culture we have in RELX around innovation and patenting is fantastic. It's not often that you have an organization where everything is in place to support you from exploration and writing to the legal aspects and patent submission process. I think RELX does great by having such a process in place.

Finding the “why” through exploration

George: I have always enjoyed exploring. When I would receive electronic toys as a kid, I would break them down to understand what was inside. I have always thrived on the scientific exploration or motivation to dive deeper into the details of things and try to make them work better. As a PhD in the field, I am very active and constantly challenge myself to take things to the next level.

Zubair: My PhD supervisor was instrumental in cultivating my exploration mindset. When I would bring him a solution, he would always say “This is great, but why this solution?” So that word, “why”, has stuck with me. We have these results, but why did we get good results, or why did we not get good results? Results are less about the numbers, and more about the why.

Patent Identifying Funding Sources with machine learning & natural language Processing

Systems and methods for extracting funder information from text

Legal language is hard for humans, and harder for machines

My team looks at the technology landscape to find the right solutions for our customers in the legal domain. We work in an area that is experiencing rapid growth in cutting-edge technologies, and it’s important to understand what's going on in the broader world to bring that into our world. Things that were not possible a few years back now seem to be more tractable. Recently, there has been an explosion in new technologies specifically in understanding written language. Many companies are trying to solve the problem, but they deal with mainstream English. We are dealing with legal language.

Legal professionals need to find relevant cases to cite, but it's hard to find cases that have very similar language to the cases they are working on. They have to run searches with lots of keywords, which isn’t easy. We created a method using deep learning and machine learning techniques that can find similarities in legal language. A year or two ago, it wasn’t really possible to solve this problem. The way legal opinions, statutes, and legislation are written, it's hard for humans to understand, so making the machine understand that language is really challenging and interesting. Now we have the right tools, techniques, and knowledge.

The evolution of creativity

When I was young, I used to take apart old electronic appliances to understand how they worked. As an undergraduate, I got into building things with friends who had similar interests, and later in my career I focused on solving problems for customers, nothing necessarily innovative; I was trying to find the best way to make things better for them. About 10 or 15 years ago, I joined the research team, and that's really what got me going. I just love to do this stuff! I am surrounded and inspired by colleagues who are very creative. Our collaboration, an understanding of our customers’ needs, and a deep understanding of the technology allow us to innovate; it’s a dream job.

Filed & Granted Patents in machine Learning

PCT/US2021/039845 (Filing), Methods and systems for performing legal brief analysis, S. Sharma, J. Punuru, M. Pendyala, & M. Shewhart

Systems and methods for generating issue networks

Systems and methods to facilitate analytics with a tagged corpus

System and methods for identifying and visualizing elements of query results

Systems and methods for creating an annotation from a document

The hard work of finding problems

Tamir: Most of the time you do things without understanding why you're doing them. You don’t stop to think how you can change something when everything is working, so the hard part is finding the problem. For example, we noticed that sometimes customers don't remember the answers to account validation questions. These are questions like, “What was your first car?” We thought of a way to use machine learning and the information we have to choose better questions that would be easier for the real customer to answer and more difficult for the fraudster. It’s this method that we have patented.

Benny: Having a knowledge of the product and of the business problems helps us actively seek potential solutions. Like with the verification questions, sometimes we need to automate and to solve problems in more modern ways than we could have 10 or 20 years ago.

Figuring out how to patent

Benny: When we managed to create something that could be patented, there was a sense of excitement and fulfillment, but the process was frustrating at the beginning. We submitted the first patent, and nothing happened for three or four years. It took five years until it was processed and approved, and we were surprised! The more recent patents we’ve worked on have only taken one or two years to process.

Tamir: We talked about patenting the innovation with our product manager because we didn’t know how to do it and discovered that there's a simple document that you need to fill out and send in to get started. Also, sometimes when you are seeking to patent an idea, other people don't think it’s a good, so you might throw it away. In some cases, the idea just needs to be more polished and tweaked, so there still can be a patent with the core idea. I think companies can increase their patents by educating their employees on the process.

Connecting reality with personality

Tamir: My father always said that from an early age I always came up with good ideas, and I guess that’s true. In the early 90s, I got a Casio watch with a touch screen, and I then I got an idea for making a touch screen keyboard so you could change everything straight on the keyboard. Now we have that technology. Someone else made it happen, and I feel a big loss there!

Benny: I have always liked arts and creative ideas. Since I met my wife, I have started enjoying it more and playing with it. I created some art projects for some festivals and came up with creative ideas for the LexisNexis Israel site. When you discover creativity, when it happens, it's more fun. It makes me view the world differently. For me, creativity connects the reality with the personality. Creativity also comes from a culture that encourages innovation and patenting.

Patent for knowledge-based authentication

Systems and methods for improving KBA identity authentication questions

A matter of timing

The patent that I was involved in was sparked by the environment and the technology that was starting to present itself. In physiotherapy, the drop-off rate of doing rehabilitation exercises is high, and it’s hard for patients to objectively measure progress themselves. Our invention is for measuring a person’s range of motion through augmented reality. It combines our revolutionary 3D technology with virtual conferencing to help assess patient mobility with accurate metrics and data through a screen. Patients and their doctors can see that they made an extra 10 degrees today. When we presented this publicly, there was great enthusiasm for the idea, and people wanted to know when the feature would be released in our product. However, the tough part of invention is the long and challenging path from the idea to delivery. We have the spark of the idea in the world, and the hardware might not be ready yet for it to be fully functioning. We're still ironing out the technical aspects of the idea, and the product will need to evolve to ensure a great market fit.

The space and time to innovate

I've always been into making things and solving problems. My very first job as a 14-year-old was building crosswords for a magazine! I thrive on the creative process in general, and I love working in the medical field. It’s wonderful to think that you are part of a team that builds products that will actually improve people's lives and patient outcomes. Pushing the boundaries and having really great people around you is so important and brings a fresh perspective when brainstorming. Some questions may sound simple, but sometimes these can be critical if we hadn’t previously considered the problem from that angle. Elsevier has given me the psychological safety to ask those questions and allowed me to take the space and time in my normal job to really focus on innovation.

Filed Patent on motion assessment

PCT/EP 2020/083237A1 (Filing), Motion assessment instrument, I. Walsh & M. Opanasiuk