Activity

savvythunder219

I have finally added a ingestion layer in the system which can take sentences and turn them into cypher and then run it in the Neo4j database. By doing this system can finally retain knowledge and move forward.

Next crucial task is adding rapid fuzzing which is a library using a string-matching technique changing similar words to known words. By adding this, the system will keep the relationships clean and sturdy.

Attachment
0
savvythunder219

Till now I have successfully added the Brain query layer, this layer is responsible for the query handling and it’s conversion to the cypher query (raw_text->cypher_query) it uses different components in order to produce a valid query.

I have done some optimization which reduces the time of execution from seconds to Mini seconds. (Hint: check the screenshot attached)

I’ll start improving the database from tomorrow and then start building a layer which actually reads data from raw text feeds the brain.

For information you can read the document, it includes everything in detail.
https://docs.google.com/document/d/1H09usmI2NbMok7FPiWfEVv_1I-ObjZTuYagHlUBOLj0/edit?usp=sharing

Attachment
0
savvythunder219

Currently working on cypher generator and Neo4j database handling.
I disposed the cypher generator and Neo4j database handler due to it’s logic misconception and it highly lacked for what it really meant to, I’m now building it very carefully and at it’s best, this will keep the project future proof and easy to maintain.

Attachment
0
savvythunder219

Nexus Devlog #2

Added NLP functionality and a Neo4j wrapper with a conflict resolver along with a entity resolver.
Now Nexus analyze queries and convert them into Cypher queries which will help it to query the database and retrieve data. I have added it for both the sides like for ingestion and egestion both.

The query is passed as a text and then decomposed into different sentences, after it is analyzed and then resolved, after that it’s Cypher(Query language for Neo4j database) is generated. This is done with a parameter ‘ingestion and egestion’ such that it can generate query for both retrieval and data input.

Next I’ll start implementing it into the main engine.

Attachment
1

Comments

savvythunder219
savvythunder219 about 2 months ago

Disposed

savvythunder219

Today I have started to implement a graph database which will be useful for LLM integration.
Till now Nexus has no memory but after the final update Nexus will be able to reason, think and remember. I’m very excited for the upcoming updates myself because I have different attachment for this project.

This dev log holds the logging of all over work till now,
Till now I have achieved a deterministic retrieval based system which works with pipelines and holds the logical power to compose it’s own response through template response and all. I have not implemented any ML yet but only regex but future updates will hold ML power and much more.

Attachment
0
savvythunder219

I’m working on my first project! This is so exciting. I can’t wait to share more updates as I build.

Attachment
0