Most sentiment analysis tools have been developed to examine simple, clear, concise communications. The trouble is, most communications from governments, corporations and other organized groups are complex, nuanced and often use veiled language. This poses a unique challenge to those seeking to automate analysis of these complex communications. Prattle has developed a methodology to solve this problem. This talk will walk through how Prattle analyzes complex language from central banks and corporations around the world. The talk will also suggest other applications for this unique methodology.
Track: Main Room
Text Analytics For A Fast Moving World
The growth and worldwide success of Twitter has surfaced natural limitations of hashtags and keyword searches. What was once a mechanism for organizing information for efficient consumption has been made impractical by overwhelming volume and diversity of discussion. Word embedding technology promises to change real-time analytics as much as it has effected static data analytics by making it possibly to follow events as they change in real time. But what does that mean, and what does big text look like? We’ll explore how using this new type of machine learning on streaming data allows for an evolving look at unfolding events with case studies of the ISIS “topic of the moment” and fear around the Zika virus outbreak. Beyond that, we’ll touch on the challenges of running text analytics for the 2014 World Cup (the biggest social media event yet) and discuss lessons learned and techniques used in order to cover the games in real time.
Identifying and Visualizing Semantic Indicators of Emerging Refugee Risks
Long before refugee crises emerge, there is a large amount of anecdotal and research data (mostly unstructured) that provides insight into devolving human conditions. The information is rarely in a single language, in a consistent form or published on well read sites. This information is very challenging to identify using traditional search techniques and structured data analytics as it is often sparse, inconsistent and not yet ‘popular’. Using data plugins that connect to more remote sites, evolving semantic models that identify increases in velocity of refugee indicators, powerful text analytics and associated visualizations, users can identify threatened regions and possible mitigating actions. This presentation discusses and demonstrates emerging refugee threat models, associated analytics and high impact visualizations in Savanna, a web based all source analysis solution.
Keynote: Cracking the Code to the Next Breakthrough in Machine Translation
Everyone is looking for the next breakthrough in machine translation. No one believes that machine translation is a completely solved problem. Most people would like to see machine translation systems produce higher quality results.
A good translation is one where the meaning of the source is preserved, and it is rendered correctly in the target language. Users expect accuracy on all of the various levels–grammar, syntax, semantics and pragmatics.
Dr. O’Neill-Brown has been carrying out research and development aimed at improving machine translation accuracy. Her work includes the development of novel methods that take advantage of semantics and pragmatics to increase user understanding of machine translated results. This talk will present the results of these research and development efforts. Dr. O’Neill-Brown will discuss how these techniques aid user understanding of machine translated results and boost machine translation quality.