Hi Graham,
Thanks for your question. I am the Optimizely engineer who has designed and mostly implemented the fulltext search in Graph.
In case you notice that fulltext is too "greedy" (e.g., too much is indexed as part of it), then you can control it by unchecking the searchable property in the CMS, as explained here: https://docs.developers.optimizely.com/platform-optimizely/v1.4.0-optimizely-graph/docs/full-text-search-support#enable-full-text-search-support-for-fields
You can also configure this in the client package by using the default annotation, as explained here: https://docs.developers.optimizely.com/platform-optimizely/v1.4.0-optimizely-graph/docs/installation-and-configuration#default-mode
When you query for the page content, do you see in the same item the content from the other pages? You can see the values of _fulltext by selecting it in your GraphQL query. If you could share the correlationId in the extensions section that you get from our API, then I can look into this with more detail. You will find the correlationId in the JSON response like this:
{
"data": {
"Content": {
"items": []
}
},
"extensions": {
"correlationId": "82f8b02b-385d-41a5-ba9a-232e037a3b75",
"cost": 21,
"costSummary": [
"Content(21) = limit(20) + fields(1)"
]
}
}
Hi Junte,
Many thanks for your reply, I have tried using the default annotation but it seemed to make no difference, just a heads up that I am not using the Optimizely Graph .Net client and am instead using the following package https://github.com/graphql-dotnet/graphql-client/tree/master - this was chosen at the time as the Optimizely client had not been released, do you suggest switching over to the Optimizely package instead?
Either way, putting the client aside, the actual Graph searches when done via the Optimizely Graph section are returning content from other linked pages, I have sent you a message on Slack with an example correlation id.
Thanks in advance.
Graham
We are currently doing semantic searches on the _fulltext field in Graph, what we are seeing however is that the content in the _fulltext field is being a bit greedy and the page content has content from other pages that are linked from that page...has anyone else experienced this issue at all, if so did you manage to find a fix??