Paper Reading #11

Comments:
Reference:
Title:A Conversational Interface to Web Automation
Authors: Tessa Lau, Julian Cerruti, Guillermo Manzato, Mateo Bengualid, Jeffrey P. Bigham, Jeffrey Nichols
Venue: UIST 2010
Summary:
This system is very similar to the previous article. Instead of querying music or photo databases, however, this interface queries web tasks. The system is titled CoCo. The commands are described as "conversational." An example given is "get road conditions for Highway 88." Also, the system uses a repository to capture previously recorded web scripts and web browsing history. The goal here is to more accurately guess how to complete a given task. This poses a problem because these scripts are not sorted by website or task.

The primary functionality is "to enable a user to direct CoCo to perform a task on the web." An example mentioned in the article was "forward phone calls to home." The command was actually displayed as being done over Twitter (hence the comic). From this, the command is mapped to actions on the web that will accomplish this task. Thus, it becomes a planning problem, which for this system relies on the use of plan libraries. Plan libraries are "user-centric repositories of scripts that describe how to accomplish tasks on the web."

Overall, this is a system designed to be easy to use, or "conversational," as the article puts it. It will assist users in completing tasks on the web easier and more efficiently by way of a simple textual interface.
Discussion:
Wow this paper was incredibly long. I did enjoy the interface idea. It seems to me like the systems and interfaces in the UIST sections have been targeted at making more simple interactions. This definitely follows under that category. It excites me to see how we're trying to make designs that are easier to use for common users or less technically savvy users. This notion reminds me a lot of The Design of Everyday Things (unfortunately). However, this paper is entirely too long. It was impossible to finish this in one sitting. It actually took me three sittings to finish this one. Reading papers with this many technical details is just very tiresome. The interface and system is a great idea in my opinion, but they need to condense their paper so as to not lose people's attention.