Engineering Problems Not Addressed

“Ask HN: What are challenges for enterprise level Knowledge Graph adoption in AI I recently got started learning about knowledge graphs, started with Neo4j, learnt about RDFs and tried implementing, but I think it requires a decent enough experience to create good ontologies.<p>I came across some tools like datawalk, falkordb, Cognee etc that help creating ontologies automatically, AI driven I believe. Are they really efficient in mapping all data to schema and automatically building the KGs? (I believe they are but havent tested, would love to read opinions from other&#x27;s experiences)<p>Apart from these, what are the &quot;gaps&quot; that are yet to be addressed between these tools and successfully adopting KGs for AI tasks at enterprise level?<p>Do these tool take care of situations like:<p>- adding new data source<p>- Incremental updates, schema evolution, and versioning<p>- Schema drift<p>- Is there any point encountered where you realized there should be an &quot;explainability&quot; layer above the graph layer?<p>- What are some &quot;engineering&quot; problems that current tools dont address, like sharding, high-availability setups, and custom indexing strategies (if at all applicable in KG databases, im pretty new, not sure)<p>- Based on your experience, which tool comes closest to accurate &quot;automated&quot; parsing or multiple data sources to KG?<p>- Also do you think applications of KG would still be relevant 5 years down the line? I think its adoption would&#x2F;is increasing but could be wrong”