Natural language processing, or NLP, is one AI-based technology that’s finding its way into a variety of verticals. We covered the business applications of NLP in our previous report, and in this report, we intend to cover the technology’s applications in finance more extensively. NLP might allow a company to garner insights that can be used to assess a creditor’s risk or gauge brand-related sentiment across the web. We researched the space to better understand where NLP comes into play in the finance industry and to answer the following questions:
- What types of NLP-based applications are currently in use in finance?
- What tangible results has NLP driven in finance?
- Are there any common trends among these innovation efforts? How could these trends affect the future of finance?
This report covers vendors offering software across three applications:
- Credit Scoring
- Sentiment Analysis
- Document Search
This article intends to provide business leaders in the finance space with an idea of what they can currently expect from NLP in their industry. We hope that this article allows business leaders in finance to garner insights they can confidently relay to their executive teams so they can make informed decisions when thinking about AI adoption. At the very least, this article intends to act as a method of reducing the time business leaders in finance spend researching NLP companies with whom they may (or may not) be interested in working.
LenddoEFL is a Singapore-based company with 115 employees. The company offers a software called The LenddoScore, which they claim can help banks and financial institutions assess an individual’s creditworthiness using NLP and machine learning.
LenddoEFL is focused on allowing financial firms in developing countries offer loans and credit services to growing middle-class populations in these countries. In most cases, these customers have little to no credit history, and LenddEFL claims its software helps banks understand lending risks based on customer digital footprints.
LenddoEFL claims users can download the Lenddo application on their smartphones, and the software uses NLP to comb through the user’s digital footprint, such as social media account use, internet browsing history, geolocation data, and other smartphone information. Then, LenddoEFL uses machine learning algorithms that convert this customer data into a credit score, which banks or credit unions can use. The system then provides their lending partners with the credit score for potential customers in their target audience.
Below is a short 3-minute video demonstrating how LenddoEFL works:
Lenddo claims to have helped FICO develop their new FICO score services in India. FICO signed an agreement with Lenddo to allow them access to digital data (with the consent of consumers). LenddoEFL was then able to develop the FICO score for and extend credit to previously unscorable consumers without increasing risk of default. There were no measurable results we could find for this partnership.
We were unable to find any mention of enterprise-level companies on LenddoEFL’s website nor in any of their press releases, but they seem to have raised a total of $14M in funding and are backed by Blumberg Capital and OMIDYAR Network.
Naveen Agnihotri is Chief Technology Officer at LenddoEFL. He holds a PhD in Neuroscience from Columbia University Vagelos College of Physicians and Surgeons. Previously, Agnihotri served as a scientist at MIT and as CTO at Milabra.
Sigmoidal is a company that offers a software which they claim can help banks by providing consultation, software development, development operations, and data tagging services using NLP and machine learning.
Sigmoidal also claims to have developed a trading software that uses machine learning to track patterns in how customers might spend, invest, or make financial decisions from their transaction history. The software then co-relates patterns in customer investment with market developments obtained by scouting news and social media to offer personalized investment advice to customers.
Sigmoidal claims investment firms can automate the task of mining for information on market developments from news sites and social media using their software, which can perform document classification and named entity recognition. Then, the software uses NLP to filter out the information that is most relevant to the investor’s needs. Sigmoidal claims that their software can also help extract details such as a person’s name and company from text in the collected data. The system then provides the collected data on a dashboard as shown below:
Sigmoidal does not make available any case studies reporting success with their software, but the company does claim to have worked in projects with NASA, DARPA, NVIDIA, Microsoft, PwC, Goldman Sachs, and Intel. That said, we could not find much public evidence of these collaborations.
Marek Bardonski is Head of Artificial Intelligence at Sigmoidal. He holds a bachelor’s degree in Computer Science from the University of Warsaw. Previously, Bardonski served as a senior deep learning research engineer at NVIDIA Switzerland for one month. That said, Bardonski is not a C-level executive at the company, and in fact, it is possible he serves an advisory role there, based on his LinkedIn profile, which lists him as an advisor for multiple companies at presents.
Marcin Mozejko is Principal Deep Learning Engineer at Sigmoidal. He holds an MS in Mathematics from the University of Warsaw, but again, he is not a C-level executive at the company.
Sentifi is a Switzerland-based company offering a software called Sentifi Maven, which they claim can help financial institutions make better business decisions by collecting insights from news, social media, financial influencers, and blogs using NLP.
Sentifi claims a business can first use Maven to identify those people on the web who are advocating for their brand the most, called influencers. Sentifi’s software then searches through the influencers’ blog posts, social media, and relevant news articles referencing the influencers. This is done using NLP. The system then provides a collection of insights from the searched media to help businesses identify topics of discussion in the market, get ideas for trading, or discover events that might impact their investments.
Below is a short 3-minute video demonstrating how Sentifi works:
Although there are instances of Sentifi using their technology to predict the impact of events that affect global financial markets, such as the Ebola outbreak, we could find no case studies in which the company worked with a financial firm in a B2B transaction.
Sentifilists UBS, Commerzbank, Vontobel, ZKB, SIX and Swisscanto as some of the company’s past clients.
Anders Bally is founder and CEo at Sentifi. He holds a PhD in Environmental Shareholder Value from the University of Freiburg. We could not verify Bally’s experience in artificial intelligence, and there does not seem to be anyone on the Sentifi team with the “data scientist” title that has a robust academic background in AI.
Nuance Communications is a Massachusetts-based company with over 8800 employees. The company offers a software called Nuance Document Finance Solution, which they claim can help financial services companies automate and digitize their documentation processes using NLP
Nuance Communications claims users can integrate their document finance solution into existing workflows without disrupting existing processes. The software uses natural language processing to automatically read and understand documents that involve loan or mortgage processing. Businesses can use their historical documentation records to train Nuance’s NLP solution. Then, the Nuance Document Finance Solution uses NLP to comb through several thousands of these documents to extract and summarize the most relevant information from them.
Nuance claims businesses can work with them to integrate the software using some amount of human input from loan or mortgage officers at the bank. The system then provides a dashboard where employees at a financial institution can access a loan or mortgage application simultaneously.
Below is a short 3-minute video demonstrating how Nuance Document Finance Solution works:
Nuance Communications claims to have helped SwedBank improve and gain insights from customer interactions. Nuance Communications claims to have developed a chatbot interface for the bank called Nina, which allowed the bank’s customers to search for information on Swedbank’s homepage search interface and get answers to basic transactional questions. According to Nuance Communications, Nina handled over 30,000 conversations per month, resolving tickets on its first response 78% of the time for the first three months after the integration.
Joumana Ghosn is Director of NLP Research at Nuance Communications. She holds a PhD in machine learning from the University of Montreal. Previously, Ghosn served as a machine learning scientist at Idilia.
Vlad Sejnoha has been the Senior Vice President and CTO at Nuance since 2001. He holds a Masters degree in Electrical Engineering from McGill University. Vlad has experience in the field of NLP and speech recognition for over 30 years and holds 22 patents to date. Vlad also heads the company’s external research relationships, including Nuance’s five-year collaboration with IBM Research.
AlphaSense is a New York-based company with 128 employees. The company offers a market data collection software, which they claim can help financial institutions create a search engine for financial market developments.
AlphaSense claims that their database of market developments are periodically indexed with additions to the existing millions of documents, such as public company filings and conference call transcripts.
The AlphaSense search engine then parses topics, concepts, and ideas from these documents to find valuable pieces of investment information. The system then provides a summary of the most relevant information for search queries from employees at financial firms on the search engine interface. Once they find information of interest, users can access more detailed information in the same interface.
Below is a short 2-minute video demonstrating how AlphaSense works:
AlphaSense does not make available any case studies reporting success with their software.
AlphaSense claims they have over 800 clients, such as banks and investment firms including 66% of Dow Jones Industrial Average (DJIA) companies.
Dmitry Kan is Head of Search at AlphaSense Inc. He holds a PhD in Computer Science from Saint Petersburg State University. Kan is also the founder and CEO at AI Startup SemanticAnalyzer.
Takeaways for Business Leaders in Finance
Credit scoring seems to be one of the more common applications for AI in finance, and vendors now are offering products that can help assess credit scores for customers with little or no credit history. Many of these products use NLP to gauge the creditworthiness of a customer from their digital footprints.
Of the companies we covered in this report, Nuance seems to have the most traction in terms of having several established use cases and white papers to go along with their products. Nuance also has several reputable NLP engineers in their organization and offer their services in a number of industry verticals, such as automotive, healthcare, and finance.
The company also seems to have developed capabilities for NLP applications such as human-like virtual assistants, extracting data from unstructured sources and improving search and discovery within organizations. The larger capability set for Nuance has also made them a robust name in the NLP space.
Another company with a healthy number of customers and established use-cases is AlphaSense who claim they have over 800 clients. Information search and discovery can be a highly viable AI solution for finance companies. This might purely be down to the act that using human analysts to comb through millions of news or social media posts to gather the most relevant developments might simply not be very scalable.
Financial institutions can expect AI vendors to offer NLP solutions for extracting data from both structured and unstructured documents with a reasonable level of accuracy. Business leaders in finance might need to be mindful of the fact that although they may have access to historical data, such as loan applications or mortgage documents, this data might be useless unless properly cleaned and tagged before can be used to train any AI models.
Header Image Credit: Jerusalem Water Undertaking