When words are “keywords”: Query design and other digital methods

Notes from the DMI Summerschool Certificate Program Day 1 by Erik Borra, Anne Helmond and Esther Weltevrede.

A lecture on query design by Anat Ben-David

Introduction

How social science usually works: Come to your object of study with pre-assigned categories. You try to fit things in. What we strive to do, in a more Latourian sense, is try to have the categories emerge (according to what the actors themselves say; to take seriously the actors’ own issue language). Key words are fundamental to proper/good/decent query design. Knowing when words are key words and knowing how to formulate and design queries is fundamental to Digital Methods.

1. Search / Research

The difference between search and research is reflected in Google trends, which is an example of mundane keywords / search.

Usually research starts with a why. Google autocomplete for why …

Google Suggest: search engine politics. How to approach a research question? Before addressing the engine. Google Suggest = a Google bomb.

Social research with the web

• Latour (1992) Program and anti-program. Issue language defined by one set of actors, also find anti-program language
• Rogers (2004) “side by sidedness” offical / non-official … different kinds of actors (living side by side) in an issue map.
• “We look at Google search results and see society, instead of Google” (Rogers et. al, 2009)

2. Words / Keywords

“Terms and Audiences.”

Specific actors use specific terms: map which actors use which terms, for example in the Occupied and Unoccupied Media Spaces project: security fence, separation wall, barrier wall, the wall, separation fence. How can one use the various terms used by various actors to map the issue of the barrier?

The building blocks of query design:

• Identify “Issue Language” among actors and in different Web-spaces
• Choose terms that are exact, yet inclusive, to include all actors’ language “side by side”
• Refine search and keywords for optimal dataset (Use, for example: AND, OR, NOT, SITE, INURL and consider language issues)
• If necessary, repeat and change query according to the results (e.g. “fish out” additional terms from returned snippet text)
• Become familiar with search operators and syntax. See, for example, http://www.googleguide.com/advanced_operators.html

Refine your queries throughout the tentative searches.

Mapping the social life of conservation

• Use a set of conservation approaches provided by experts
• Design queries of the terms to make them exact and inclusive
• Extract conservation place names from snippets of text and turn into subsequent keywords
• Query all discovered conservation actors in …

How to turn specific terminology into keywords? The words are not specifically related to conservation. Query design: ”ecosystem approach” + conservation

Test out what type of results you get with different type of query designs, before you start scraping.

A source set is a list of actors, sometimes the urls returned by search engine results.

Program and Anti-Program

“~Cellular Phone” AND “brain tumor” AND “not associated”
“~Cellular Phone” AND “brain tumor” AND “270%”
Compare two queries across different actors. Add “site:.edu”, “site:.com”, etc

The tilde (~) indicates synonyms are allowed for in the results.

Program and Anti-Program (or: “search engine politics” AND “do the math”)

How is the issue of “google street view” and privacy being treated when google-related sites are excluded from the search?

1. “Google street view” +privacy
2. “Google street view” +privacy site:google.*
3. “Google street view” +privacy -site:google.*

Before and After

• Keywords as the starting point of the research (ScrapeGoogle tool). When do keywords enter the research protocol? Start with keywords as the main identifyer of an issue space.
• Keywords as a subsequent analysis of a demarcated space or network (Lippmannian Device, Open Calais, Issue Discovery Tool)