Search vs filter

One of the big decisions I am working on for this project is deciding which keywords are for filtering and which will just be for search. 

My preliminary list has over 130 descriptors. That's way too many to work well for filters, but filtering is part of the essential plan for the site. 

So, I need to sort the keywords into groups, decide which are important to use in filters, which might be extraneous, and which we should still use in the metadata for searchability. I'm narrowing down which I think need to be used in the filters with help from staff via card sorting.

I also have started sketching a very preliminary wireframe, those famed boxes and arrows, to explain how I'm thinking about the design. It's helpful to have this when I explain what I want done with my index cards.

Preliminary pencil sketch

What I'm finding as I do this is that it's actually quite hard to group some of these concepts and determining which are important isn't always clearcut. 

In my first pass through the index cards, I created a group I called "purpose" which contained concepts related to how a type of technology might be used. Included in this group were keywords like "research", "preservation", and "virtual tour". 

Now, those might be terms that are useful as filters or not. They may be better left to search. But a conversation my supervisor had today with a potential client highlights how having those types of keywords could be valuable. The people she was talking to wanted to be able to create virtual tours of the library and the different spaces there. Since a virtual tour could be done in VR, panographic images, 360 images, and maybe other options, being able to search and find previous projects that are virtual tours would be helpful. 

So, I'm still sorting and resorting keywords, gathering data, and really thinking about how they might be used. Hopefully I'll have a clear idea by the end of next week so I can move on to designing the interface.

Week four: Descriptors and keywords

My focus this week is descriptors. Descriptors are the words used to describe the various assets, the words that may eventually become keywords. 

I say may because, for this project, not all of the descriptors might be useful as keywords. Some will be in the text description while others will be used as metadata on the items. It's really the ones that end up in metadata that are of particular concern. 

Deciding which facets are important to be able to search on is part of what my job is on this project. So, while there are many ways to describe the digital assets, not all are useful for searching, sorting, or filtering. It's my job, along with my supervisor and the manager, to understand which are useful and which aren't. 

Currently, I have 139 descriptors. I'm still conducting interviews so the list may grow, although each interview elicits fewer new terms. I have started standardizing some of the vocabulary (the original list of terms was over 150). One task I need to start is categorizing the descriptors. With such a long list, they need to be organized into facets or they will be too unwieldy to use. 

Tried and true

My idea is to use card sorting, which is often used to organize the hierarchy of a website, to sort the descriptors into facets. Using knowledgeable staff to help determine what category different descriptors belong to and find good labels will be much more productive than me trying to figure out whether a term should be, say, "technology" or "tool". I'm a fan of using physical cards, especially with a small group of people doing the sort, although there are online tools I could use. Looking into those tools to see if there is one that would suit my needs is on my plan for Thursday.

In classes, I've done some card sorting (in Z515, Information Architecture) and I've created a faceted vocabulary (in Z503, Representation and Organization), but those were mostly theoretical applications or with familiar domains. Actually taking a bunch of information from an unfamiliar domain and organizing it is a little intimidating but also a good test. And, I don't have to do it all myself. That's the point of having the staff participate in a card sort: they will help guide the final organization. They and clients (users) will need to understand the organization and structure, so it needs to be based in already established organization.