The prevalence of Slack, and its ubiquity in the office had me wonder, about ways that we could improve the horribly poor search performance of Slack. Often times communication is messy, and retrieval of message history given high amounts of volume creates a horrible road block to get to relevant results. What , then is required to make this better. I theorize that the gateway to improved search experience, is not in keyword relevance, but in understanding a query. Search could be improved 10x , if you could ask a bot to simply search for relevant search results translating phrases that have to do with the topic at hand. Really you probably don’t always want to see the last 5 times an item was mentioned, sometimes you want to find information that must be derived . The following is an attempt to step through how Slack search in particular could be improved.
Search on slack is a black hole for the office communication. Content goes in and if you need it you simply scroll up slack history until you find the statement you were looking for. Alternatively you filter history a little bit searching for snippets of phrases that you might need . One task that we often do is look for people. sometimes when a person is temporarily out office or sick, the most convenient way to signal this is to simply drop a message in Slack. This is fantastic, because its easy, this is horrible because its not really searchable.
My purpose is to create an @slackbot that answers the query where is
This is no different than asking google questions, its really just an extension of the principle to a specific domain, where the volume will be lower, and the chances of success much higher, as the domain won’t be as broad. Businesses have lots of needs like this, the point of interest is that slack/txt/gchat is really a place that could benefit from a richer context aware search.