Bloom not finding node

[andy_hegedus]

Hi,
I have a database where one of the nodes uses a node key.
around word. In Bloom, when I enter a term - "leakage" which is a know node, Bloom finds it properly as a node.
Now I have a second database that uses very similar schema especially around the word nodes.

In Bloom, when I do search for a known node, it does not find the nodes.

I can validate in browser that the nodes exist.

The indexes appear very similar.
The original database that works as expected

and the problematic database

What should I check to get the same behavior where I can enter a term of the word node into the search field in Bloom and get the appropriate nodes returned.
Andy

[jens_oknelid]

Hi Andy!

Are the indexes on the word.term property constraints or property indexes? It's a bit hard to see from your screenshot but the name "constraint_term" suggests that it might be a constraint and not an index.

If it is a constraint, Bloom will not search that property unless the label (word in this case) has less than 1000 nodes. So could it be that there are <1000 term nodes in the "works as expected" DB but more then 1000 in the problematic one?

[andy_hegedus]

Hi Jens,

The one that works has 572 nodes and the one that doesn't has 15639 so your hypothesis is consistent with that.
Question how do I work around that? Do I create a full text index? The nodes have two properties the term and "part of speech" (pos) and I really only need to search around the term property.
Andy

[jens_oknelid]

You can create a property index for the properties you want to be able to search. In your case the query would be CREATE INDEX FOR (w:word) ON (w.term) (for Neo4j >= 4.1, for older it's slightly different). It's possible to add more properties after w.term separated by commas.

[andy_hegedus]

Hi Jens,
Thank you very much. It now works as expected. The 1000 node limit was really causing confusion. I did the same as last time and got different result.
Here is a graph I can create. The blue nodes are words and the yellow documents. In this case you see the multiple parts of speech of a word and the major ones here are verb and noun. You see documents that only use the word as verb, another cluster where it is used as a noun, and the middle where both parts of speech are used.

Andy

[jens_oknelid]

Great! Nice to see that you found the node styling, not too many people discover that 🙂

[andy_hegedus]

Jens,
A question about node styling. Can I save it?
Use case: I am working on a database and create a set of styling. Then I expand the database by adding some new node types and relationships. For Bloom I need to recreate the perspective to see these nodes and relationships which resets all the styling including the styling rules. So it would be nice to be able reload the node/relationship styling and then only have to adjust the newly added node and relationship types.
Andy

[jens_oknelid]

Hi again,

When adding new labels, you should be able to create new categories in the perspective for the new node types you added (by opening the perspective sidebar drawer and clicking the "add category" button). Then, you will be able to keep your styling information.

In general, it's possible to export the perspective, which includes the styling information, to a file that you can import later (for example if you want to transfer a perspective from one DB to the other). This is done from the little dropdown on the perspective cards in the perspective dashboard popup.