Patty's update from the Geospatial Computational Social Sciences conference at Stanford

Patty Frontiera from the D-Lab went to the Geospatial Computational Social Sciences conference at Stanford on Monday 10/20 (https://css-center.stanford.edu/geospatial-computational-social-science-conference).

Here is her summary:

1. GIS for Exploratory Data Analysis

• The presentations showed that geospatial analysis and mapping using desktop GIS, R and python are an extremely important part of exploratory data analyses in the social sciences.

2. GIS for Communication / Visualization

• Digital maps and web maps are an important part of communicating the results of scientific analysis. However effective communication with any kind of graphic / visualization tool may require additional funding for design professionals which researchers typically are not.

3. Garbage in Garbage out - or good computational tools don't replace good thinking.

• Ed Chi a research scientist at Google gave a great talk on the analysis of implicit location data in twitter content. One point he made is that there are great tools for parsing and analyzing these data but bad data can creep in when the tools are used without thoughtful consideration of the data inputs and outputs. For example, just because someone entered "Donkey-Kong, Texas" as their home town and a geocoder parsed that and returned a valid coordinate pair does not mean that that town exists. He would be a great speaker to get at Berkeley.

4. Uses of GIS in Social Science

• A panel discussed the uses of GIS in social science research. The key points they made were:
  • GIS is an important tool for linking social and environmental data.
  • GIS is important for exploring data at different geographic and temporal scales.
  • The use of GIS in social science research requires and benefits from an interdisciplinary approach.

5. Academia-Industry Collaboration

• There were three industry speakers, one each from Facebook, Google, and EBay. They discussed collaboration with academia, making the following points:
  • Because it takes so long to establish a working relationship and because a tremendous amount of effort goes into creating data sets that can be made available for social science research, universities should work on developing long term relationships with industry rather than come ask for data for one-off projects.
  • Academia should participate in the development of open standards for space-time geospatial data formats.
  • Academia should not insist on overly restrictive licensing terms.
  • Industry likes collaborating with social scientists (as opposed to computer scientists) because they have different goals from industry and thus it is more of a mutually beneficial rather than competitive relationship.

6. Social Science Research Needs with regard to Geospatial computational Analysis:

1. Social scientists need training in the following areas: GIS, R, Python, SQL, data cleaning, geospatial data file formats, and computing infrastructure.
2. Data reuse and research replication: because of how long it takes to obtain and clean data, social scientists need infrastructure to facilitate data sharing and reuse.
3. Academia needs to recognize the value of data intensive social science though the use of alt metrics. Stanford Press just received a Mellon grant to explore alt metrics movement.

7. Social scientist..data scientist..computer scientist?

• There was a heated discussion on whether or not a social scientist needs to become a computer scientist and what the nature of the relationship should be between these two fields.  This was a really good discussion which may be worth having at Berkeley.
  • Do social scientists need to become data scientists?
  • What level of computational training is enough?
  • Do computer scientists need social scientists too?
  • There is a tension in the disciplines of applied computer science (maker culture) and social sciences/humanities (idea culture).