You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Original search PR: #984
Related issue: #1024 (analytics)
Midway through development, it was noted that the relevance sorting of search results may be sub-optimal for some particular searches. For example, searching for "sickness policy", you need to scroll down to the fourth result to find the correct page
We'll need to have a think about/research whether this is a symptom of the search itself or the way things are structured. As mentioned in #1024, we could look to tweak the structure of content for internal SEO, or we could look into how to boost certain results for particular search queries
The search engine, Elasticlunr.js, has a "query-time boosting" feature, which we may be able to tap into to boost (or maybe deboost?) certain terms. For example, if "policy" is combined with another word, we might prioritise the other word since "policy" is likely more generic. I believe the engine already weights words based on how frequently they appear across the entire dataset/corpus, but boosting may be a way of augmenting this
The text was updated successfully, but these errors were encountered:
Original search PR: #984
Related issue: #1024 (analytics)
Midway through development, it was noted that the relevance sorting of search results may be sub-optimal for some particular searches. For example, searching for "sickness policy", you need to scroll down to the fourth result to find the correct page
We'll need to have a think about/research whether this is a symptom of the search itself or the way things are structured. As mentioned in #1024, we could look to tweak the structure of content for internal SEO, or we could look into how to boost certain results for particular search queries
The search engine, Elasticlunr.js, has a "query-time boosting" feature, which we may be able to tap into to boost (or maybe deboost?) certain terms. For example, if "policy" is combined with another word, we might prioritise the other word since "policy" is likely more generic. I believe the engine already weights words based on how frequently they appear across the entire dataset/corpus, but boosting may be a way of augmenting this
The text was updated successfully, but these errors were encountered: