Decision Conversations Decoded     

links

 photoKillian Levacher photo

Decision Conversations Decoded - overview


Decision Conversations Decoded*

 

The purpose of our project is to help people when they make a decision collaboratively, i.e., involving more than one person.

Many decisions in our lives we cannot make alone. Sometimes we have to work with our family, friends, colleagues, or professionals to find the best option. And it is no secret that these collaborative discussions can be inefficient, disorganized and lead to decisions that leave everyone frustrated.

One challenge with collaborative decision discussions is that we get lost in the exchanges. We cannot both participate to the conversation and keep carefully track of which alternatives were proposed, by whom, when, what criteria were brought forward, how do they relate to the options proposed, who brought up the criteria, who thought it was important and so on.

The objective of our project is ti build a system and a set of algorithms that can do decision-oriented note taking and support us during our decision conversations, reminding us of where we are, and after those conversations, when we need to reflect on what has been discussed, sometimes to reach a final decision.

We are in the process of develpoing such system and algorithms. The core of the technology lies in the information extraction algorithms needed to identify the relevant elements(alternative, criteria, decision topic). In addition we are interested in detecting assent and groiup agreement along with targeted sentiment of speakers towards each element. 

 

*We sometime refer to the system informally as DECODE for short.




Decision Conversations Decoded system

At this point, we have an initial prototype of our system.

You can access our
Decision Conversations Decoded system (Click Here)
using credentials demo/demo

(There is a "Demo Text" button in the interface that can be used to populate some sample text into the input text box)


Decision Gisting Corpus

We are also making the annotated corpus available for people interested in alternative and criteria extraction.
Technical Report about Annotation
Folder with Data and ReadMe file