Text Analytics Forum 2020 Conference Program

The Text Analytics Forum invites all who deal with text to take a deep dive into this powerful set of techniques. The Forum has something for all: whether you are new to the field and want to understand how it can add new capabilities or you are an experienced text analyst and want to see what the latest techniques and tools can add to your repertoire. At the Text Analytics Forum, you’ll be able to share ideas and network with peers and get practical advice and thought leadership from experts in the field. You’ll learn everything; how to get started, how to make the business case for text analytics, the latest developments and best practices in the field, and use cases showcasing the cutting edge of myriad applications. Check out the full program below.



Tuesday, Nov 17

Optional Conference Day

 

Optional Conference Day

08:00 AM2020-11-172020-11-17

Tuesday, November 17: 8:00 a.m. - 5:00 p.m.

Upgrade to a Platinum or Gold Pass for full-access to all sessions at KMWorldEnterprise Search & Discovery, and Taxonomy Boot Camp, a series of co-located events happening alongside Text Analytics Forum 2020. Also includes access to the Networking Happy Hour taking place in the Enterprise Solutions Showcase from 5:00 p.m. - 6:00 p.m.

 

Networking Happy Hour in the Enterprise Solutions Showcase

05:00 PM2020-11-172020-11-17

Tuesday, November 17: 5:00 p.m. - 6:00 p.m.

Wednesday, Nov 18

Keynotes: Wednesday

 

Keynote - The Disrupted Mindset & Knowledge Sharing

08:30 AM2020-11-182020-11-18

Wednesday, November 18: 8:30 a.m. - 9:15 a.m.

Growth is always hard, and disruptive growth is exponentially harder. It requires companies to make tough decisions in the face of daunting uncertainties.  Some organizations beat the odds and succeed at becoming disruptive: Adobe, ING Bank, Nokia, Southern New Hampshire University, and T-Mobile, among them. Their stories make it clear that organizations don’t have to be tech start-ups or have the latest innovations to transform. What they need to do is develop a disruptive mindset that permeates every aspect of the organization. Li lays out how to do so by focusing on three elements:

  • a strategy designed to meet the needs of future customers;
  • leadership that creates a movement to drive and sustain transformation; and
  • a culture that thrives on disruptive change.

Drawing on interviews with some of the most audacious people driving disruptive transformation today, Li will inspire leaders at all levels to answer the call to lead disruptive transformation in their organizations, communities, and society.

Speaker:

, Senior Fellow, Prophet and Co-Author, Groundswell: Winning in a World Transformed by Social Technologies

 

Welcome and Keynote: A Deep Text Look at Text Analytics

10:45 AM2020-11-182020-11-18

Wednesday, November 18: 10:45 a.m. - 11:30 a.m.

What are the current and future trends for the field of text analytics? Join program chair Tom Reamy for an overview of the conference themes and highlights and a look at current and future trends in text analytics. This year’s theme is how to utilize the structure in so-called unstructured content to build highly accurate text analytics applications. Also, in addition to the Ask the Experts panel, we’re adding a panel on AI and text analytics. We’ve reorganized the conference into two tracks, Enterprise and Social, to better reflect how the two diverge and converge. Last year’s hot topic was knowledge graphs; what will this year’s be?

Speaker:

, Chief Knowledge Architect, KAPS Group, LLC

 

Wednesday, Nov 18

Track 1, Wednesday: Social Text Analytics

 

Text, Subjectivity, Sentiment, and Emotion

11:45 AM2020-11-182020-11-18

Wednesday, November 18: 11:45 a.m. - 12:30 p.m.

Text isn't your typical data source. Unlike fielded transactional and operational data, text—social, online, and spoken—captures both facts and feelings, conveying sentiment, opinion, intent, and emotion. Sentiment analysis and emotion AI technologies apply machine learning to extract both hard data—entities, aspects, events, relationships—and subjective evaluations from text sources. In recent years, we've seen sentiment, emotion, and intent analysis uptake for consumer, healthcare, and diverse other fields implemented via conversational interfaces, data science tools, and customer intelligence and market research applications. Presenter Seth Grimes draws on broad industry and technology experience to survey sentiment, emotion, and intent analysis research, technology, applications, and the marketplace for text and speech. He presents case studies and explores ethical considerations to provide a comprehensive update on text subjectivity, sentiment, and emotion understanding.

Speaker:

, Principal Consultant, Alta Plana Corporation

 

Emotion Recognition: Sentiment Meets Psychology

01:30 PM2020-11-182020-11-18

Wednesday, November 18: 1:30 p.m. - 2:15 p.m.

Everyone understands intuitively what emotions are. We all know how they feel to us and how others appear to feel them—yet large-scale research on how humans experience and express emotions remains difficult. Given that human language reflects the human experience, text analysis is an ideal way of accessing emotion. Bag-of-words approaches are highly effective for gauging a wide range of psychological attributes but fall somewhat short when measuring emotion. More modern methods, such as those that utilize deep neural nets, can fall short for different reasons. By contrast, a rule-based system that accounts for grammatical complexity provides significant benefit over other approaches for capturing emotion. We explore the psychological differences between the familiar concept of sentiment and the more detailed concept of emotion. We then explore the pros and cons of available emotion analysis methods using a text-based dataset that captures the reported emotions of 2,500 individuals during the COVID-19 crisis. Finally, we explore how the same methods can be applied to a range of applications, from mental health management to customer service improvement in call centers and advertising optimization during esports matches.

Speaker:

, Head of Science, Receptiviti

 

AI and Combating Bias

02:30 PM2020-11-182020-11-18

Wednesday, November 18: 2:30 p.m. - 3:15 p.m.

Debiasing AI-Powered Text Analytics Applications
2:30 p.m. - 3:15 p.m.

One of the most-discussed topics in AI has been the growing realization that AI-based systems absorb human biases and prejudices from training data. While this has only become a hot news topic within the last couple of years, organizations building AI-powered text analytics applications are working on new solutions to address this issue. In this talk, Speer discusses the following:

  • What it means when we talk about bias in text analytics and how someone really decides what bias in text analytics is
  • The business implications of having biased analysis of text
  • How a company organization that wants to tackle this issue can figure out whether or not its text analytics system is biased
  • What Speer is doing to remove bias from text analytics systems at Luminoso
  • How other companies can apply what Speer has done to their own systems to make them less biased
Speaker:

, Chief Science Officer and Co-Founder, Luminoso

Detecting Bias and Prejudice Online: It's Not Just Black and White
2:30 p.m. - 3:15 p.m.

Determining fake and real news is becoming increasingly difficult. Historically, resources, credentials, and accountability were challenges in becoming a reporter or a source for providing information. With the internet, social media, and anonymity, any person (or bot) can create and disseminate news and information. This venue that allows the quick and free dissemination of information continues to create an environment of unreliable, inaccurate, and insidious influencers. There is a psychological tendency to believe that any published word is true and accurate. There is no easy way to distinguish whether what you are reading is the original source or accurate. Our presentation identifies challenges and an automated methodology for ranking the credibility of information as an extension of Hovland's classic persuasion research through the use of a number of classifiers. 

Speakers:

, Attorney, PRX Solutions, LLC

, Partner, PRX Solutions, LLC

 

Ask the Experts Panel

04:00 PM2020-11-182020-11-18

Wednesday, November 18: 4:00 p.m. - 5:00 p.m.

A panel of four text analytics experts answers questions that have been gathered before the conference, during the conference, and some additional questions from the program chair. This was one of our most popular features last year, so come prepared with your favorite questions and be ready to learn.

Speakers:

, CEO & Founder, Smartlogic, USA

, Expert System Enterprise

, Regional Vice President of Sales, Lucidworks

, VP Engineering and Founder, Voise, Inc.

 

Wednesday, Nov 18

Track 2, Wednesday: Enterprise Text Analytics

 

Not Just What, But Where: Leveraging Document Structure to Improve Relevance in Automated Tagging

11:45 AM2020-11-182020-11-18

Wednesday, November 18: 11:45 a.m. - 12:30 p.m.

Human annotators look at zones in a document, for example, the title, abstract, or lead paragraph, in order to identify signals for aboutness. Text analytics applications follow a similar process. This session looks at the impact of configuring text extraction software to analyze or weigh document sections selectively. We look at ways to improve relevance scoring by leveraging the structure of documents and see how analyzing text within a structural context can lead to a better determination of aboutness and, ultimately, better end use of the data.

Speaker:

, Corporate Taxonomist, IBM

 

Auto-Tagging

01:30 PM2020-11-182020-11-18

Wednesday, November 18: 1:30 p.m. - 2:15 p.m.

Utilizing the Structure and Context in Technical Standards to Improve Auto-Tagging
1:30 p.m. - 2:15 p.m.

Technical standards and specifications, as a genre, have their own unique internal patterns and language. In this interesting case study, we describe using this structure and language to improve autocategorization. The first task was to leverage XML structure to consistently chunk up the technical standards. We wanted to take advantage of the inherent recursive containers, or sections, of the standard to improve auto-tagging accuracy. The ultimate goal is tying semantic tags to the precise section of the standard in which the software found evidence for them. The second part of the presentation focuses on also using autocategorization software to capitalize on terms in hierarchical paths in a taxonomy in order to differentiate between homographs in the taxonomy.  

Speakers:

, Ontology Architect, National Fire Protection Association

, Director, Product Research and Planning, Bloomberg BNA

Testing Autocategorization Output
1:30 p.m. - 2:15 p.m.

If you're auto-tagging documents and other textual artifacts, how do you know if the tags are accurate? How do you know if the accuracy is improving? Testing and reviewing auto-assigned tags is one of the most important ways of improving an autocategorization system as well as one of the most time-consuming. But testing requires a substantial number of sample documents to produce an accurate result set for reviewing the results. Creating a strategy for testing the autocategorization system that includes organizing the test results will allow you to fine-tune and maintain your autocategorization system across time. This presentation focuses on creating a repeatable testing process that can be used throughout the lifecycle of the autocategorization system. Participants learn about setting baselines for tagging accuracy, choosing sample documents, maintaining records of results, reviewing those results, and managing an ongoing testing strategy. 

Speaker:

, Taxonomist, Office of Justice Programs, Leidos

 

Finding: Data Triage and Semantic Technology

02:30 PM2020-11-182020-11-18

Wednesday, November 18: 2:30 p.m. - 3:15 p.m.

Best Practices for Data Triage of Multilingual Text: Ingestion, Analytics, and Visualization
2:30 p.m. - 3:15 p.m.

Efficient data triage means finding the text you need fast. This talk lays out best practices for collecting and triaging data by applying the most recent natural language processing (NLP) technologies. Traditional approaches of machine translating multilingual text to English and then applying NLP can lead to mangled proper nouns and translations and hinder true understanding. We review data triage best practices with an ingestion pipeline of Twitter data and analyze each step of the process to illustrate practices that you can use within your organization. Touchpoints include data ingestion, NLP, categorization, search, and data visualization. To find both novel and relevant documents from a clustered set of records, we discuss grouping your data to surface a diverse set of the most salient documents to provide the highest value to the analyst or end user.

Speaker:

, Director, Field Operations, Basis Technology Corp

Semantic Technology Accelerates Retrieval of Relevant Text
2:30 p.m. - 3:15 p.m.

The number of subject matter experts available to authoritatively construct reliable information pales in comparison to the continual onslaught of data to triage. Linguists and human translators are overwhelmed by the sheer volume and variety of documents to translate. Which ones should they translate first? Numerous medical studies are published on an ongoing basis, but doctors only have a handful of minutes to read daily before tending to patients. Which documents should they read and what sections are most relevant? Semantic technologies, most notably latent semantic indexing (LSI) and text embeddings, have found extremely successful niches in both intelligence and legal discovery. Not only can they find a needle in a haystack, they can also go beyond, rapidly identifying the most relevant needles in a needlestack. We take a look at best practices for semantic search, how it can be used for classification, and see how it can be applied to solve real-world problems today.

Speaker:

, Federal Solutions Engineer, Basis Technology

 

Ask the Experts Panel

04:00 PM2020-11-182020-11-18

Wednesday, November 18: 4:00 p.m. - 5:00 p.m.

A panel of four text analytics experts answers questions that have been gathered before the conference, during the conference, and some additional questions from the program chair. This was one of our most popular features last year, so come prepared with your favorite questions and be ready to learn.

Speakers:

, CEO & Founder, Smartlogic, USA

, Expert System Enterprise

, Regional Vice President of Sales, Lucidworks

, VP Engineering and Founder, Voise, Inc.

Thursday, Nov 19

Keynotes: Thursday

 

Keynote - AI Transformation & Knowledge Flows

08:30 AM2020-11-192020-11-19

Thursday, November 19: 8:30 a.m. - 9:15 a.m.

Join Marco Iansiti as he shares insights on the revolutionary impact AI has on operations, strategy, and competition beginning with a look at the core of the new firm, a decision factory he calls the “AI factory.” All the more relevant in the age of COVID, the AI factory is where analytics systematically convert internal and external data into predictions, insights, and choices, which in turn guide and automate operational workflows.  “As digital networks and algorithms are woven into the fabric of firms, industries begin to function differently and the lines between them blur. The changes extend well beyond born-digital firms, as more-traditional organizations, confronted by new rivals, move toward AI-based models too,” says our speaker.  Gather ideas to meet the challenges of a new “reset” world and find the correct strategies to harness AI for your organization.

Speaker:

, Professor of Business Administration, Harvard Business School and Coauthor, Competing in the Age of AI: Strategy and Leadership When Algorithms and Networks Run the World

 

Thursday, Nov 19

Track 1, Thursday AM: Social Text Analytics

 

Analyzing Spoken Conversations for Real-Time Decision Support in Mission-Critical Applications

10:15 AM2020-11-192020-11-19

Thursday, November 19: 10:15 a.m. - 11:00 a.m.

This talk covers the work we do to analyze spoken conversations between customers and CRM or call center agents in mission-critical applications. We discuss the additional challenges of making sure your speech-to-text technology can deal with your domain concepts. We also look at how text analytics becomes even more complicated, but also more fun, when you can never be sure who said what and if what was said was correctly transcribed.

Speaker:

, CEO, Franz Inc.

 

Making Experience Text Analytics Actionable: The Human/AI Balance

11:15 AM2020-11-192020-11-19

Thursday, November 19: 11:15 a.m. - 12:00 p.m.

If you're not analyzing your customer speech and text, you're getting an incomplete picture, which could be misleading your decisions. Furthermore, discovering the top topics from unstructured data is certainly important, but has anyone noticed that top topics alone are not actionable? It takes strong human intervention to align text topic categories to your business decision areas. Additionally, a great text analysis should be living/breathing—ensuring it stays relevant as your business and markets change. In this presentation, we help you understand what steps you need to take to make your text analytics truly actionable according to your own stakeholders.

Speakers:

, Director, Analytics, Confirmit

, Principal Text Analytics Consultant, Confirmit

 

KMWorld Awards and Attendee Luncheon Keynote

12:00 PM2020-11-192020-11-19

Thursday, November 19: 12:00 p.m. - 1:00 p.m.

 

Thursday, Nov 19

Track 2, Thursday AM: Enterprise Text Analytics

 

Six Markets and Millions of Documents: Strategies for Developing a Global Taxonomy

10:15 AM2020-11-192020-11-19

Thursday, November 19: 10:15 a.m. - 11:00 a.m.

Indeed is a job search engine that serves more than 60 countries and indexes millions of documents. It strives to go beyond keyword search to enable users to search by concepts relevant to their job interests—such as skills, licenses, and benefits. This presentation is about how Indeed generated a global taxonomy by extracting custom concepts across multiple languages and markets. Over the past 2 years, we developed a taxonomy of concepts designed for the U.S. market and a system for extracting them. Replicating the process that created these resources for the U.S. was not scalable to the international markets, as the number of market and language experts required would have been too high. We used a multi-pronged iterative approach that spreads the risk and responsibilities across several solutions: rule-based models, text mining, clustering, and supervised models. Each of these solutions required a human in the loop for annotation, curation, and subject matter expertise. Attendees walk away with an understanding of the different entity extraction options and how to use them for taxonomy creation. They also learn about the cross-team and human-machine collaboration required to implement such an approach.

Speakers:

, Taxonomy Analyst, Indeed

, Product Scientist, Indeed

 

Topical Analysis and Auto-Summarization of Scientific Content

11:15 AM2020-11-192020-11-19

Thursday, November 19: 11:15 a.m. - 12:00 p.m.

A key need of our users, researchers in academia and industry, as well as our internal editorial colleagues, is dealing with information overload. Most critically, it is making sure they are aware of all relevant information for their research or the focus of their publication. To address these needs, Springer Nature partnered with Digital Science to focus on text analysis applications which assist users in three tasks: capturing all published research relevant to their topic of interest, providing analytics to explore trends and collaborative networks of those topics, and applying auto-summarization solutions to provide summaries in various formats. This presentation briefly reviews the key user needs before providing details of the analytics applications and the types of output we have achieved.

Speaker:

, Director, Business Development, Springer Nature

 

KMWorld Awards and Attendee Luncheon Keynote

12:00 PM2020-11-192020-11-19

Thursday, November 19: 12:00 p.m. - 1:00 p.m.

Thursday, Nov 19

Thursday PM

 

AI in Action

01:00 PM2020-11-192020-11-19

Thursday, November 19: 1:00 p.m. - 1:45 p.m.

Text Analytics and Artificial Intelligence for Government
1:00 p.m. - 1:45 p.m.

Every day, international government agencies face tremendous challenges. They must ensure a safe, livable society; make informed spending decisions; and regulate a healthy economy. The volume of data that supports these missions is exploding, and it is increasingly unstructured. While manually reading through hundreds of documents may be manageable, reviewing thousands or millions of documents is not. This presentation showcases the application of text analytics and visualizations to transform unstructured, unmanageable data into action. This involves demystifying AI and machine learning in text analytics. We show how to apply these techniques to use cases, such as assessing information from the COVID-19 pandemic, adverse events, situational awareness, and regulations analysis, to find answers in minutes versus months. Many of these cases reference a more technical paper ,which is publicly available to the group. We also discuss evolving best practices in text analytics strategies. 

Speaker:

, Principal Solutions Architect, SAS

Deep Learning to Embrace Your Jargon
1:00 p.m. - 1:45 p.m.

The Inter-American Development Bank is a multilateral development institution with a mission to work with governments and other actors to address development challenges in Latin America and the Caribbean. This mission motivates the Bank to be constantly strengthening its capacity by generating and acquiring knowledge from its operations in 26 countries, as well as from external sources. Fundamental to this task has been the creation of structure from unstructured data, specifically in leveraging deep learning approaches to create a dynamic, and constantly updated model of our organization-specific jargon and use of language, in a multilingual environment, and building applications that use the model to understand the nuances of user queries and deliver the most relevant results. So far, we've used this approach to map employee knowledge within the organization and to power-search for people within the organization. This talk discusses the process of developing the model and reflects on trying multiple technology approaches and future applications. 

Speaker:

, Senior Knowledge Management Specialist, Knowledge, Innovation, and Communication Sector, Inter-American Development Bank

 

New Text Analytics Techniques

02:00 PM2020-11-192020-11-19

Thursday, November 19: 2:00 p.m. - 2:45 p.m.

Text Analytics and Graph Databases: A Powerful Combination
2:00 p.m. - 2:45 p.m.

Graph databases offer a powerful way to enhance content, model complex relationships, and organize both structured and unstructured information. As part of his presentation, Hilger shares best practices around how to use graph databases to enhance the power of text analytics. He also shares two different use cases where he helped clients increase the power of their content using a combination of text analytics and graph databases. At the end of the presentation, attendees understand new and interesting ways to combine text analytics and graph databases to improve the way information is used throughout their organizations.

Speaker:

, COO and Co-founder, Enterprise Knowledge LLC

‘How Did We Miss That?’ Detecting and Responding to Weak Signals at the Working Level
2:00 p.m. - 2:45 p.m.

Traditional project management tends to be rearview mirror-oriented. Large volumes of data are often filtered, summarized, and time-lagged, with little insight into the many small events at the activity level that collectively impact overall performance. As such, critical alerts are often generated far downstream from root causes and conditions. Problems that could be quickly and easily addressed are allowed to grow, sometimes until it's too late. This is significant, as correcting errors and re-work can reach up to 80% or more of the total life cycle cost of a system. This presentation describes a next-generation project management system that combines human and machine intelligence in ways that result in more efficient and effective problem identification, prevention, and resolution. Initial test results based on historical data for several high-visibility projects will be presented, including the use of text analytics to detect weak signals present in project discourse which point to potentially serious problems downstream and to link those weak signals to historical data, lessons-learned, and other knowledge in order to generate recommendations/solutions. 

Speaker:

, CEO, Applied Knowledge Sciences, Inc.

 

Closing Panel – AI & TA

03:00 PM2020-11-192020-11-19

Thursday, November 19: 3:00 p.m. - 3:45 p.m.

AI has seen a string of spectacular successes in recent years in multiple industries. However, most of those successes have been in the realm of data. The AI track record in text has been more uneven. Nevertheless, AI has become an important technology in text analytics. Join us for an interactive session that discusses the latest AI techniques, the breakthroughs, and issues, as well as the critical obstacles that unstructured text presents to AI. We also discuss the important role that text analytics plays in the world of AI.

Speakers:

, Chief Science Officer and Co-Founder, Luminoso

, Principal Solutions Architect, SAS

Thursday, Nov 19

Keynotes

 

Closing Keynote - The Role of Knowledge & Information in Crisis Management

04:15 PM2020-11-192020-11-19

Thursday, November 19: 4:15 p.m. - 5:00 p.m.

Crisis management has moved from planning to a day to day reality however organizations are ill equipped to manage a situation where we are dealing with unknown unknowables or have to deal with multiple Black Elephants (something that changes everything!) competing for resources and attention.  What is the role of knowledge and information in a crisis?  How do we gain attention to weak signals where anticipatory actions would reduce downstream risk and increase overall resilience. Shifting from Just-in-time to Just-in-case sounds like a good idea but it is far from simple and in a resource starved environment may simply not be possible.   For the last few decades we have based practice in industry and government on an engineering metaphor, focusing on efficiency. This approach is, to quote Lincoln, “Inadequate to the stormy present.”  Are there better approaches that we can adopt by treating the organization and society as a complex ecology?  Would such a metaphor shift allow us to do more with less? Last year’s conference ended with a rousing discussion of creating resilience in organizations and society. They discussed transforming and revolutionizing the way we do business as we move into an uncertain future, how we satisfy our clients in an ever-changing technological age, and how, in our complex societies, we provide value, exchange knowledge, innovate, grow and support our world. Our popular, and sometimes controversial, speaker Dave Snowden has again assembled a group of experienced thinkers and doers who are capable of reimagining a future based on uncertainty.

Speaker:

, Chief Scientific Officer, Cognitive Edge

Co-Located With