Are the possibilities for converting unstructured text into queries on objects?
For example the user might leave a note that they need a certain feature of an object that might or might not exist. How to then search the universe of objects for that feature? Like if we are talking about cars, the request might be asking for a particular make which might be a feature but they also might be asking for it to be manufactured on the moon which is not in the set of features we track. Or they might be asking for a feature that is similar described in a text property of an object.
Example car object:
Make: BMW
Color: Red
Manufacture: 1997
Comments: “Drives fast but turn signal does not work”
Potential queries:
Queries that i would like to return:
- “BMW”
- “Cars with broken blinkers”
Other possible queries:
- “electric vehicle”
- “german cars” ← this one would be cool but a nice-to-have
Thank you!
It seems like you’re trying to convert a single object’s properties to “free text queries” so that you can find other similar objects in the Ontology that have a matching property. If that isn’t what you’re trying to do, can you add a bit more detail about your usecase. If ^ is your goal.. in Logic, you can ask the LLM a question and provide it with some context. The context can be either input passed as an argument or data that your LLM queries from your Ontology.
If you’re looking to produce free-text queries from a single object (a Car), you can pass the Car instance into the useLLM board and prompt it to return an Array containing a list of queries generated from the object’s properties.
If you’re looking to produce queries from all objects of a given Object Type (all the cars) stored in your ontology, you can give the useLLM board access to the Query Object Tool. This will allow it to produce queries after querying the Ontology for sample values.