During the readings of this week, I learned that there is actually a lot of complexity behind the faceted search features that I find are ubiquitous in most of my information seeking activities. Especially on e-commerce websites, and through IUCAT, I use faceted search all the time to efficiently narrow down results in a way that saves time and helps me find relevant information. I didn’t realize that behind all of those facets, there are very complex backend challenges that exist (Tunkelang). Some of those concerns are scale, efficiency, and information overload. Faceted search is computationally challenging, and can require the distribution of data across many servers when concerned with large data sources. Creating efficient and useable facets requires controlled vocabularies and metadata that makes information searchable beyond straight text-search. This is a daunting task for people managing data, especially when there seems to be infinite information that needs tagging. Indexing of documents can be inefficient and complicated. Metadata tagging cannot always be completely automated, and often requires humans to clarify nuanced categories. Crowdsourcing of information is an option to determine metadata tags from user behavior and user-created categories. In addition to the back end problems, there are also front-end issues of how to display many facets, and how to determine which ones are critical to the user’s needs. So yeah, I won’t take faceted search for granted anymore, and I will appreciate when websites like e-commerce allow me to use conventional ways of narrowing down choices to the items that meet my information needs.
Taking Facets for Granted
14 Mar This entry was written by thschone and published on March 14, 2015 at 7:50 pm. It’s filed under Uncategorized.