Tuesday, December 1, 2009

Relevance: An Easy Search, Unpacked


I don't speak Spanish, but can usually figure out the meaning of traveler's information printed in a couple of languages. My insufficiency became evident when searching online for information on La Posada, a Mexican Christmas tradition.

I wanted to know if there was traditional music for this Mexican celebration. I think I would have found it right away if there weren't other meanings and spellings for 'la posada.'

Let me unpack this search. I find it helpful to deconstruct search experiences to see what works and why.

I started with what I (thought I) knew: the name of the event is La Posada. I remember producing a musical a number of years ago with this title, so figured it would make good keywords.

My first query was: la posada
I tend to favor sparse queries, since every word creates constraints that may result in unexpected results.

Scanning the first page of results, it became apparent that la posada has other meanings. I ended up with a page of hotels, retirement communities, bungalows and lounges located mostly in the southwest and Texas.

Armed with this information, I knew I needed to add something specifically about music or Christmas to the query, so next I tried: la posada music. I hoped this combination would be unique to the activity I was looking for.

Now the results were relevant. There were enough keywords in the snippets related to my topic to make me think I was on the right trail: song, carols, lyrics, young children... These are keywords I think should be associated with the event.

Not only did I find a traditional song for La Posada, the results also provided a clue: there is another way to spell la posada: las posadas. I didn't need that to find the music, but it could have come in handy.

Thinking about the skills needed for this search, here are a few observations:

  1. You have to start with what you know or what you're given
  2. It's OK to start simple (my two word query)
  3. You have to read and interpret search results to determine if your simple query was effective. That requires enough knowledge to know whether information is relevant or not. I suspect this could be a very deep rabbit hole to go down, since (to me) relevance depends on language skills and the making of meaning. With sufficient experience, however, most of us are able to tell that information about hotels and restaurants does not match information about a traditional Mexican Christmas practice.
  4. You have to know associated keywords to tell if you are on the right track. I uncovered a lot of synonyms for music and those signaled to me I was getting close.
Working with younger students, I would use this search to give them an authentic experience to think about words they encounter in snippets that tell them they are getting warmer or colder. A simple question may be all it takes when looking at snippets: "How can we tell if we are getting warmer (closer) or colder?" Or, "what words do you see that show you are getting warmer or colder?"  Have the students name the keywords.

This may seem so elementary and resemble a language arts lesson rather than a search lesson. In fact, better searching could be a secondary goal.

Telling if results are relevant or not can be difficult. What if you encounter a bunch of words you don't know? Sometimes these turn out to be better keywords than you started with, but it's similar to using a foreign language. Taking time to look closely at all the keywords in snippets really can help you search better.

By the way, a single word search for 'posada' is even more interesting. Try this with students and see what they think is relevant.

No comments: