411 implement combimatch to improve reasoner performance #526
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Look at this later, because is this how we want it?
Had to introduce new strategy that limits the full matching to a single rule. This was neccesary to make meta-knowledge-interactions work faster. Also had to remove some invert match code, which I found strange, but had to do to make it work. There are still quite some tests the fail, so those need to be fixed first. Starting with the looping test.
Included matching configuration using Enums and mapped those to 4 levels of interoperability. Knowledge Gaps are not yet working, so the TestDynamicSemanticComposition test is not yet working. Also, the matching algorithm needs to also support the highest levels.
However, it does not fully work yet. Some tests are still failing and we need to compare whether the results are comparable with the previous graph pattern matcher. Also, maybe discuss the different levels of interoperability with others and make sure the TKE fits into that. Extra configuration options to add to TKE (as opposed to reasoner): - do not include meta graph patterns (ReasonerProcessor) - only go one level deep with reasoning (ReasoningPlan#createOrGetReasoningNode)
Things still to do: - replace matcher with reasoner? and remove matcher? - more testing on performance?
Todo: - fix unit test post react.
This configuration options allows the user to configure whether the meta KIs of the Knowledge Engine should be included when executing a particular user KI. Also removed the SerialMatchingProcessor entirely and always use the ReasonerProcessor.
- correctly dealing with splitted forward reasoning. - make biggest matches not contain too many matches (which are actually not biggest).
Using default Java Collections parallelism really improves the performances and makes it utilize multiple CPU cores much better. Although it is not enough, because there are just too many bindings that need to be merged.
411-implement-combimatch-to-improve-reasoner-performance Also added a debugging mechanism to help finding out why the reasoner does not return the info you expect.
Mainly: - took over error messages from serialmatchingprocessor in reasoning processor. - also include binding set validation in Java API, instead of only in REST API.
Although that is not necessarily caused by the new matching algorithm.
bnouwt
deleted the
411-implement-combimatch-to-improve-reasoner-performance
branch
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.