Track 1: New Text Analytics Applications

Track 2: Text Analytics in Practice: How To

Monday, September 30, 2013

7:30-8:45am
Room: Commonwealth Hall

Registration & Networking Breakfast


8:45-8:50am
Room: Cambridge 1

Conference Chair Welcome Remarks

Speaker: Tom Reamy, Chief Knowledge Architect, KAPS Group

[ Top of this page ] [ Agenda overview ]


8:50-9:30am
Room: Cambridge 1

Keynote
Going with Stop Words to Reveal Personality and Social Processes

The smallest and most forgettable words in any language are pronouns (e.g., I, we, it), articles (a, the), prepositions (to, of, for), and a handful of other function word categories. Over the last several years, research in social psychology has demonstrated how people’s natural use of function words in natural conversation, literature, emails, blogs, and tweets can reveal their motivations, honesty, status, , relationships with others and even sentiment. Implications of using stop words to track and understand human behavior will be discussed.

Speaker: James Pennebaker, Professor and Chair, Department of Psychology, University of Texas at Austin

[ Top of this page ] [ Agenda overview ]


9:30-10:00am
Room: Cambridge 1

Mini-Workshop: Quick Start for Text Analytics

While text analytics solutions offer the hope of beginning to tame the growing information overload, few organizations have the experience to confidently assess what text analytics is, what it can do for their business, or how to select the right tools and right methodology to get the most out of text analytics. Without this experience, organizations can either miss out on a powerful new tool or worse pick the wrong tool and the wrong implementation strategy and waste their time and resources without solving their information overload problems.

This mini-workshop presents a methodology for getting started with text analytics that ensures that organizations will get the most out of text analytics. It consists of three elements. First, an audit that assesses the organization’s content and content structure, the information behaviors and needs, and their people and technology. This produces a text analytics strategy. Second, a text analytics software evaluation process that helps the organization navigate the ever changing vendor landscape. Third, an initial pilot project that creates the foundation for future text analytics applications development, including initial taxonomy development or assessment and training of internal resources.

We will end with a group exercise knowledge audit.

[ Top of this page ] [ Agenda overview ]


10:00-10:25am
Room: Commonwealth Hall

Exhibits & Morning Coffee Break

[ Top of this page ] [ Agenda overview ]


10:30-11:15am
Room: Cambridge 1

 All LevelsTrack 1: New Text Analytics Applications
Topic: Intelligence Applications

Analyzing K- 12 Disciplinary Records (If I Had to Do It Again)

In the world of public education, disciplinary problems have grown in number and severity. Using text analytics, it was possible to look into developing solutions to these problems by looking into unstructured narratives describing the incident. The unstructured data describing the infraction yielded the true source of the problem. This session would focus on what was done, and what should have been done to produce better results that would have been accepted more readily by administrators.

Speaker: Chris Bowman, President, Educational Analytics and Logistics

[ Top of this page ] [ Agenda overview ]


10:30-11:15am
Room: Cambridge 2

 Expert/Practitioners levelTrack 2: Text Analytics in Practice: How To
Topic: To Be Announced

A Case Study of Text And Data Analytics at a Canadian Telecommunications Company

Many Canadian Telecommunications companies want an integrated way to gain additional insight into their customer base through the combination of call center agent notes and direct customer survey results with the traditional customer marketing metrics. The proposed session would give; a representation of the architecture and process to integrate the multiple big data sources; a demonstration of the discovery and automation process for classification (both categorization and sentiment); a simulated analyst workbench to consume the results of the process.

Speakers:
Tim Trussel, SAS

[ Top of this page ] [ Agenda overview ]


11:20-12:05pm
Room: Cambridge 1

 All LevelsTrack 1: New Text Analytics Applications
Topic: Enterprise Applications

Text Mining of Electronic Health Records

The ability to extract information from EHR could benefit doctors, nurses, and insurance specialists. However, the analysis of EHR data is a daunting task: information is dispersed through piles of textual documents, while the semantics of healthcare processes and professional jargon are involved.

Text mining automates extraction of relevant features from EHR data. The extraction and intuitive presentation of current symptoms, progression of problems, suitable ICD-9 codes, family history, test results, prescribed medications, side effects, allergies, and other custom features of interest enables healthcare professionals quickly grasp the holistic picture of disease and patient thus enhancing the quality of care.

Speakers:
Sergei Ananyan, CEO, Megaputer Intelligence

Edward Hoffer, Associate Clinical Professor of Medicine, Harvard and Assistant Director, Laboratory of Computer Science, Massachusetts General Hospital

[ Top of this page ] [ Agenda overview ]


11:20-12:05pm
Room: Cambridge 2

 Expert/Practitioners levelTrack 2: Text Analytics in Practice: How To
Topic: Knowledge Organization

Organizing Text and Excavating Value from Organizing Systems?

Lynda Moulton shares observations about what has worked for client vocabulary management programs. The hurdle is where to begin when the pile of terminology for any enterprise is huge and growing. Organizations launching business analytics, sentiment analysis, or enterprise search initiatives will learn about:

  • Initiating a vocabulary framework
  • Considering tools for managing terminology
  • Finding sources of specialized, unique and proprietary terms for an enterprise within a vertical business domain
  • Harvesting candidate terms
  • Maintaining, sustaining and refining lists

To be successful, the effort must be sustained and sustainable.

Speaker: Lynda Moulton, LWM Technology Services

[ Top of this page ] [ Agenda overview ]


12:05-1:00pm
Room: Commonwealth Hall

Lunch

[ Top of this page ] [ Agenda overview ]


1:00-1:45pm
Room: Cambridge 1

 All LevelsTrack 1: New Text Analytics Applications
Topic: Social Media

Social Graph Analytics: Text Analytics Approaches for Affinity Applications

Organizations are attempting to leverage social media and social connections through LinkedIn profile data, and Facebook affinities. Other social application network attributes to develop tools and applications to discover hidden relationships or desirable connections for ecommerce, networking, talent development, career planning, search enhancement, content curation, expertise location and even knowledge management applications. The core to any of these systems is the same – describe attributes, normalize the data, develop matching engines, run clustering algorithms and execute queries across large data sets. In this session we will discuss best practice for placing social networks in your unstructured data strategy.

Speaker: Seth Earley, CEO, Earley & Associates

[ Top of this page ] [ Agenda overview ]


1:00-1:45pm
Room: Cambridge 2

 Expert/Practitioners levelTrack 2: Text Analytics in Practice: How To
Topic: Enterprise Applications

Phrase Analytics: A New Dimension in Text Analysis

Organizations spend large amounts of money documenting using text models. Without analytics, there is little to show for these efforts.

Analysis confirms that changes need to be made in the structure and direction of an enterprise but it is not clear where to start and end the efforts. With limited resources, the change effort must be targeted to improve the bottom line results.

Phrase analytics is a key management technique for analyzing change and assessing the implications of decision-making. Analyzing phases used to describe the business, management better understands the impact of change and assess various business alternatives.

Speakers:
Frank Kowalkowski, President, Knowledge Consultants

Gil Laware, President, Information By Design

[ Top of this page ] [ Agenda overview ]


1:50-2:35pm
Room: Cambridge 1

 All LevelsTrack 1: New Text Analytics Applications
Topic: Big Data

Marrying Structured Data and Unstructured Text, An Analysis of the National Highway Traffic Safety Administration

Using publicly available data that has the interesting characteristic of being both well structured and having significant unstructured text, I will discuss various ways to approach this sort of analysis, using the NHTSA data as the primary source.

Speakers:
Kevin Reale, Regional Director, Analytics 8
Jeff Catlin, CEO, Lexalytics

[ Top of this page ] [ Agenda overview ]


1:50-2:35pm
Room: Cambridge 2

 Expert/Practitioners levelTrack 2: Text Analytics in Practice: How To
Topic: Big Data

Overcoming Text Analytics Barriers: Selecting the Right Tool for the Job

While there is great value locked deep in companies’ textual assets, mining this information can be both time-consuming and expensive. These costs often serve as a barrier to entry, preventing companies from capitalizing on the business value inherent in text. Fortunately, there are many tools available which can help you begin to transform this unstructured data into actionable intelligence. Through case study examples, this session will explore multiple technologies, including MapReduce, which can be used to tackle many typical text mining problems, help you discover the possibilities buried in your text and boost your business case.

Speaker: Janine Johnson, Director of Analytics, Verisk Analytics

[ Top of this page ] [ Agenda overview ]


2:35-3:00pm
Room: Commonwealth Hall

Exhibits & Afternoon Break

[ Top of this page ] [ Agenda overview ]


3:00-3:45pm
Room: Cambridge 1

 All LevelsTrack 1: New Text Analytics Applications
Topic: Enterprise Applications

MFIP “Financial Data Factory Model”

George Roth will present a new data factory concept that was implemented in the US and became an Industry Utility for the Mutual Fund Industry. He will present the challenges of using NLP to extract financial data and maintain the provenance information.

Speaker: George Roth, CEO, Recognos

[ Top of this page ] [ Agenda overview ]


3:00-3:45pm
Room: Cambridge 2

 Expert/Practitioners levelTrack 2: Text Analytics in Practice: How To
Topic: To Be Announced

The Infernal Enterprise and the Sentiment of Semantics

With an enterprise over half a century old consisting of domains as far reaching as the cosmos, defining the granularity and quality of user expectations is an infernal exercise. At NASA’s Johnson Space Center we are challenged by a multitude of variables impacting the information ideal. Text analytics, by-way-of iterative semantic enhancement, provides a ferry through the processes of knowledge capture, development, and transfer for foreseeable needs and unanticipated future use. Operational Excellence Program Manager David Meza and JSC Taxonomist Sarah Berndt review a 5 year path of semantic development and its perceived application.

Speakers:
Sarah Berndt, Taxonomist, Johnson Space Center
David Meza, Operational Excellence Program Manager, Johnson Space Center

[ Top of this page ] [ Agenda overview ]


3:50-4:35pm
Room: Cambridge 1

 All LevelsTrack 1: New Text Analytics Applications
Topic: To Be Announced

Automating Expertise Identification Using Information You Already Have

The Knowledge Systems organization at Sandia National Laboratories has developed a system that can identify organizational and staff expertise from multiple, existing, free-text information sources and present results in an interactive display. A key advantage of this approach is that it is self-maintaining. The system uses information already in our environment, and is kept current through normal work processes. The system will allow us to better respond to external inquiries of capabilities. It will enable identification of internal collaboration partners and reduce duplication of effort. It will help us know what expertise we are losing due to retirements.

Speaker: John Mareda, Manager, Knowledge Systems, Sandia National Laboratories

[ Top of this page ] [ Agenda overview ]


3:50-4:35pm
Room: Cambridge 2

 Expert/Practitioners levelTrack 2: Text Analytics in Practice: How To
Topic: Knowledge Organization

Evaluating the Quality of Auto-Categorization Results

“Autocategorization” is “auto” only in part–there is much in the process that still requires old-fashioned human judgment. One critical step the humans must take is to evaluate the quality of results and feed refinements back into the process. But how do you measure quality when human indexers themselves apply topics inconsistently and often differ over applicability of topics? This case study will consider one publisher’s experience with quality measurement in light of human variability, and will offer thoughts on what is “good enough. “

Speaker: Larry Lempert, Director, Product Research and Planning, Bloomberg BNA

[ Top of this page ] [ Agenda overview ]



4:35-4:45pm

Vendor Elevator Pitches

GNIP

[ Top of this page ] [ Agenda overview ]


4:45-5:45pm
Room: Cambridge 1

EXPERT PANEL

Session description is coming soon!

Speakers:
Lynda Moulton, LWM Technology Services

Paul Hofmann, CTO, Saffron Technology
Andrew Fast, Chief Scientist, Elder Research Inc.

[ Top of this page ] [ Agenda overview ]



5:45-7:00pm
Room: Commonwealth Hall

Networking Reception

[ Top of this page ] [ Agenda overview ]


Track 1: New Techniques and Approaches in Text Analytics

Track 2: Text Analytics Case Studies

Tuesday, October 1, 2013

8:00-8:45am
Room: Commonwealth Hall

Registration & Networking Breakfast


8:45-8:50am
Room: Cambridge 1

Conference Chair Welcome Remarks

Speaker: Tom Reamy, Chief Knowledge Architect, KAPS Group


8:50-9:35am
Room: Cambridge 1

Keynote
Future Directions in Text Analytics

What is the future for text analytics? Will it continue to be dominated by text mining applications which offer the advantage of automatic processing? Will the Social Media boom go bust or settle into a mature but hardly revolutionary market? Will the market for text analytics continue to grow 20% a year or will it crash or take off? Will there be revolutionary new techniques/applications or will we see a slower evolutionary growth?

To answer any of these questions, we have to understand what is driving text analytics and what the primary obstacles are. Long time text analytics speaker and practitioner, Tom Reamy, takes a look at a number of themes that will likely shape the near future of text analytics.

The first is how to add intelligence, depth, and meaning to text mining methods. This involves a set of strategies to dynamically combine what cognitive scientists call Fast (text mining) and Slow (text analytics) Thinking which the human brain does all the time. . This also involves new approaches such as looking at different kinds of text such as function words that our keynote speaker, James Pennebaker is exploring.

The second theme is how to build on the Watson-like ensemble methods. Is the IBM approach like Watson for Healthcare, the right or only way or are there cheaper and smarter ways to develop Watson-like capabilities for the rest of us?

A third theme is how to make text analytics easier to develop. Right now every project is a custom job with no or very little resources to build upon and requires a great deal of development by either taxonomists/linguists and/or programmers/mathematicians. Is there a way to reduce this cost by developing libraries of resources and/or making TA software much, much easier to use? Or perhaps something really new like a learning platform for text analytics software?

Join us for a speculative look at the future of text analytics – and bring your own ideas to the discussion.

Speaker: Tom Reamy, Chief Knowledge Architect, KAPS Group

[ Top of this page ] [ Agenda overview ]


9:35-10:05am
Room: Cambridge 1

Mini-Workshop: Strategies for Choosing your Next Text Analytics Project

In today’s economy, getting a strong return on investment is an absolute necessity for a new analytics project. However, with so many project opportunities, it can be difficult to choose the right one. In this mini-wokshop, we share lessons learned and strategies for success from our experience with over 200 analytics projects. Tips include advice on the timing of projects, building the right team, and selecting the right technology.

Speaker: Andrew Fast, Elder Research Inc.

[ Top of this page ] [ Agenda overview ]


10:05-10:35am
Room: Commonwealth Hall

Exhibits & Morning Coffee Break

[ Top of this page ] [ Agenda overview ]

10:40-11:25am
Room: Cambridge 1

 All LevelsTrack 1: New Techniques and Approaches in Text Analytics
Topic: Big Data

Enhancing Search with Predictive Analytics

Search is the cornerstone of enterprise information management. As the amount of searchable information grows, however, keyword relevance becomes increasingly inadequate for finding valuable information. In this talk, we highlight successes we’ve had incorporating results from predictive analytics models into search, including predictive facets and enhanced relevance functions.

Speaker: Andrew Fast, Elder Research Inc.

[ Top of this page ] [ Agenda overview ]


10:40-11:25am
Room: Cambridge 2

 Expert/Practitioners levelTrack 2: Text Analytics Case Studies
Topic: Enterprise Applications

Discovering Brand Impact from Text

Congratulations, you’ve had a video go viral! Or condolences – a product launch didn’t go the way you anticipated. Now what? What will be the long term brand impact and did your social media or consumer event even matter?

We’ll demo how consumer-facing companies evaluate the impact of high traffic point events on brands and products as well as how lessons from such events can be used on more every day occurrences.

Speaker: Catherine Havasi, CEO, Luminoso

[ Top of this page ] [ Agenda overview ]


11:30-12:15pm
Room: Cambridge 1

 All LevelsTrack 1: New Techniques and Approaches in Text Analytics
Topic: Knowledge Management

Taxonomies for Auto-Tagging Unstructured Content

When people think of taxonomies, they think of structure. But even unstructured content can make use of taxonomies to facilitate search, discovery, and content findability. This session gives a brief introduction to taxonomies, an overview of the different auto-indexing technologies of information extraction and auto-categorization, and explains the different ways that taxonomies can support auto-indexing. Case examples of rules and synonyms that support auto-indexing of unstructured content will also be presented.

Speaker: Heather Hedden, Taxonomy Consultant

[ Top of this page ] [ Agenda overview ]


11:30-12:15pm
Room: Cambridge 2

 Expert/Practitioners levelTrack 2: Text Analytics Case Studies
Topic: Intelligence Applications

From Obstacle to Opportunity: The Value of Text Analytics for BI

Social media data, customer service records and feedback emails all contain invaluable information about what customers think and how they experience your company and products. As companies gain access to these channels, text analytics tools are being recognized for their value in deriving critical insight from information previously locked in inaccessible unstructured information.

Semantics’ understanding of the meaning of words brings an unprecedented level of precision and recall to the extraction of information and the discovery of patterns and relationships. Learn how to leverage this powerful tool to derive insight from all of your information assets.

Speaker: Bryan Bell, Executive Vice President, Expert System

[ Top of this page ] [ Agenda overview ]


12:20-1:15pm
Room: Commonwealth Hall

Lunch

[ Top of this page ] [ Agenda overview ]


1:15-2:00pm
Room: Cambridge 1

 All LevelsTrack 1: New Techniques and Approaches in Text Analytics
Topic: Knowledge Organization

A Systemic Artificial Intelligence Approach to Difficult Text Analytics Tasks

Text analytics for sparse unstructured data or data sets with initial poor quality, for example social media posts or user-created product information, pose many hard challenges and is often approached in a number of experimental ways in order to reach an acceptable solution. By viewing problems in this class as solvable using a multitude of highly specific classifiers, feature extractors and predictive models that directs the control flow, complete systems can be efficiently realized that produces strong results while providing robustness and allows for incremental improvements.

Speaker: Lars Hard, CTO, Expertmaker

[ Top of this page ] [ Agenda overview ]


1:15-2:00pm
Room: Cambridge 2

 Expert/Practitioners levelTrack 2: Text Analytics Case Studies
Topic: Knowledge Organization

Semantic Publishing

What is common between the BBC, Press Association, Euromoney and all Dutch daily newspapers? They all chose non-disruptive semantic layers to push forward their business. In the last five years, a combination of linked data based enrichment and dynamic publishing were the major innovation driver in authoring, editorial and publishing aspects of content production. Over the years, a repetitive solution pattern for semantic publishing got mastered and applied over and over again. We will tell you how it works and how to achieve it – from the domain aware schema modeling, through sophisticated text analytics, to readers’ profiles aware publishing.

Speaker: Georgi Georgiev, Head of Text Analytic Division, Ontotext AD

[ Top of this page ] [ Agenda overview ]



2:05-2:50pm
Room: Cambridge 1

 All LevelsTrack 1: New Techniques and Approaches in Text Analytics
Topic: Big Data

Cognitive Computing With Associative Memories

We all have heard of IBM Watson Cognitive Computing, which works the way humans learn and reason. There is another commercial cognitive computing player, Saffron Technology.

Saffron’s Associative Memory combines semantics and statistics for advanced search and prediction on top of a super fast triple store.

Saffron delivers a real-time, always learning cognitive computing platform that is used for example for predictive maintenance and spare parts management at Boeing, for threat scoring at The Bill and Melinda Gates Foundation, to disrupt drug trafficking at JAITF, and for automated ECG diagnoses at a top hospital in New York.

Speaker: Paul Hofmann, CTO, Saffron Technology

[ Top of this page ] [ Agenda overview ]


2:05-2:50pm
Room: Cambridge 2

 Expert/Practitioners levelTrack 2: Text Analytics Case Studies
Topic: Knowledge Organization

Semantics and Semantics: Integrating Text Analytics and Ontologies

Two key enablers – and often one OR the other – come to mind when ‘semantic’ tools are mentioned: text analytics tools and conceptual models (taxonomies, thesauri and ontologies.) In many enterprise platforms, they are separate capabilities that occasionally work together. They can be more tightly integrated in a mutually beneficial arrangement. The ontologies can provide valuable inputs to the text processing pipeline, and the processing pipeline can add valuable model information to the ontologies. In this session, we will describe how we’ve achieved this integration with great success, reference architecture included!

Speaker: Christine Connors, Chief Ontologist, Knowledgent

[ Top of this page ] [ Agenda overview ]

 


2:50-3:10pm
Room: Commonwealth Hall

Exhibits & Afternoon Break

[ Top of this page ] [ Agenda overview ]


3:10-3:55pm
Room: Cambridge 1

 All LevelsTrack 1: New Techniques and Approaches in Text Analytics
Topic: Enterprise Applications

Novel Use Cases for Text Analytics

Rajesh Radhakrishnan of the University of Virginia will discuss a set of novel or inventive use cases for applying text analytics, including web, semantic, affinity analytics among others, for addressing IT problems.

Speaker: Rajesh Radhakrishnan, Principal Consultant and Chief Architect, Zygous

[ Top of this page ] [ Agenda overview ]


3:10-3:55pm
Room: Cambridge 2

 Expert/Practitioners levelTrack 2: Text Analytics Case Studies
Topic: Big Data

Content Intelligence: 4 Case Studies in the New Way of Working Using Text Analytics

This session uses four client case studies to examine the value of text analytics and illustrate the new way of working required by modern information management (aka Big Data).

The case studies will illustrate:

  • Monetization of content in media
  • Predictive analytics in healthcare
  • Knowledge recycling in research
  • Contract compliance in finance

Speaker: Gavin Green, Senior Business Consultant, Smartlogic

[ Top of this page ] [ Agenda overview ]


4:00-5:00pm
Room: Cambridge 1

EXPERT PANEL

Session description is coming soon!

Speakers:
Bryan Bell, Executive Vice President, Expert System

Tim Trussel, SAS
Catherine Havasi, CEO, Luminoso
Christine Connors, Chief Ontologist, Knowledgent

[ Top of this page ] [ Agenda overview ]


5:00-5:05pm
Room: Cambridge 1

Conference Chair Closing Remarks

Speaker: Tom Reamy, Chief Knowledge Architect, KAPS Group

[ Top of this page ] [ Agenda overview ]