Thesis Proposal: Kerry Chang
When
-
Where
GHC 4405
Description
THESIS DEFENSE
A Spreadsheet Model to Create Data-Driven Applications Using Online Data
Kerry Shih-Ping Chang
THESIS COMMITTEE:
Brad A. Myers (Chair), HCII
John Zimmerman, HCII and Design
Niki Kittur, HCII
Margaret M. Burnett, CS, Oregon State
DOCUMENT:
http://www.cs.cmu.edu/~shihpinc/Thesis_proposal_Kerry_Chang.pdf
ABSTRACT:
The Internet is full of all kinds of open data. Being able to easily use these data in custom ways could benefit many individuals. For example, a house buyer may want an application that takes a location and finds the average property price using a real-estate web service along with crime information in a CSV file downloaded from data.gov. Another example might be a restaurant manager who wants an application that regularly collects the restaurant’s reviews using Yelp’s web service and visualizes the correlations between the restaurant’s weekly sales and the average rating on Yelp. Creating such applications currently requires significant programming to retrieve data from web databases, manipulate the collected data into desired forms, and create dynamic user interfaces to present the results.
My dissertation presents a spreadsheet tool called Gneiss that makes contributions in significantly reducing the barriers to using online data and for creating data-driven applications. Gneiss will include techniques to dynamically retrieve or stream data from a variety of data sources such as REST web services, web pages and mobile sensors, and to use the data in a spreadsheet without writing conventional code. It extends spreadsheets to handle structured data, streaming data and to facilitate data exploration in spreadsheets by introducing new spreadsheet functions and interaction techniques. Moreover, Gneiss allows the user to not only create visualizations of spreadsheet data but also to program interactive web applications that can dynamically present or modify spreadsheet data using only the spreadsheet language. With Gneiss, the user can create web applications that use data from multiple web sources or turn a spreadsheet into a database to store user input data from a web page.
Gneiss targets end-users who are knowledgeable in spreadsheet programming. It supports all the above features using a consistent equation-based evaluation model familiar to spreadsheet users without the need for event‐based programing. As with regular spreadsheets, Gneiss is a live programming tool and achieves a “programming-with-example” style as users develop programs with visible example values.
In this talk, I will discuss related work that motivates the design of Gneiss, present the current Gneiss system and the proposed work, which includes several extensions to support spreadsheet data exploration and creating mobile web applications, and a plan for evaluation to refine and validate Gneiss’s features.