By Xavier Beauchamp-Tremblay
As some of you may have noticed, an improved search was deployed on CanLII last week. Here’s a summary of what it does:
1. The search engine can tell if you’re looking for a specific document or if you’re doing legal research.
Searching for a specific document is a very different behaviour than doing legal research without knowing which documents will provide you with the answers you need. The updated version of our search engine is better at distinguishing between the two.
If you were looking for a specific document, for example our usual suspect the Supreme Court decisions in Dunsmuir v New Brunswick, a search for “Dunsmuir” (in either one of the first two search boxes) search wouldn’t return Dunsmuir first. The new engine understand these types of queries better than before and will return Dunsmuir first.
The same was true when searching for specific statutes, but now, when you search “Criminal Code” (again, in either one of the first two search fields), the Criminal Code will be the first hit.
The system also does a lot of matching of relevant metadata under the hood, so searches for common acronyms of statutes, for instance PIPEDA, will return the right document, among other similar improvements.
2. It taps into the wealth of commentary available on CanLII.
Until earlier this year when we added law reviews (and after that many other sources of content), CanLII had only a limited number of texts in its commentary section, and the search algorithm was optimized for primary law. This meant that it rarely returned commentary in the top results. Now it does, assuming of course that we have a piece of commentary that is retrieved with the keywords you entered.
A search for “Gladue” will return commentary in the 3rd and 4th rank (at the time of this post): https://www.canlii.org/fr/#search/text=gladue
3. Results are displayed in a way that makes it transparent how they match your query.
To help the users analyze the search results and understand why certain documents were returned, the keywords now appear in bold in the title of the document in the list of results. The same is true for the references.
The search engine will also adapt how it presents the results to you depending on the keywords you have entered. For instance, if using a parallel citation like one from the Ontario Reports (O.R.), the search will adjust and display that reference in the results… as opposed to the neutral citation which would have been displayed had you, for instance, searched the document using the name of the parties involved.
If you look for the Rizzo & Rizzo Shoes case by searching “Rizzo shoes”, the results won’t display the less often used “O.R.” (Ontario Reports) citation:
But if you specifically search using the Ontario Reports reference, the O.R. citation will be displayed: https://www.canlii.org/en/#search/text=36%20OR%20418
It’s a bit more technical, but the highlighting feature in the document itself now highlights the references.
We hope you like these improvements. Feel free to give us feedback.
This is the second improvement to the search engine this year. To know more about both the previous update and get a glimpse of the future, you can read this post by Marc-André Morrisette, the maestro behind our search engine.
Solex also stands for SolrCloud Lexum plugins, the latest iteration of the search engine Lexum deploys in all of its products.