Needing help with Knowledge Graph Index..

[bilalkhan1011]

Hi,

I have been working on trying to solve this issue for days on end now but can't seem to get it to work. I am trying to build a KnowledgeGraphIndex and then save, and load it in my application. This seems to work sometimes but my tool will fail to find any matching information when I query. When I try to turn my loaded index into a networkx graph it also shows as completely empty which feels like it should not happen.

This is how my index is built and stored:

graph_store = SimpleGraphStore()
storage_context = StorageContext.from_defaults(graph_store=graph_store)
root_id= "kgindex"
index = KnowledgeGraphIndex.from_documents( documents=documents,
max_triplets_per_chunk=2,
storage_context=storage_context,
embed_model=embed_model,
include_embeddings=True,
show_progress=True)
index.set_index_id(root_id)
storage_context.persist(persist_dir="./index")

and this is how I'm loading it in which doesn't work for some reason:

root_id = "kgindex"
storage_dir = "./index"
graph_store = SimpleGraphStore()
storage_context = StorageContext.from_defaults(graph_store=graph_store, persist_dir=storage_dir)
index = (load_index_from_storage(storage_context))
query_engine = index.as_query_engine(similarity_top_k=2)

Any help would be massively appreciated as I have been trying for so long.

[dosubot[bot]]

Hey there, @bilalkhan1011! I'm here to assist you with any bugs, questions, or contributions you may have. Let's tackle this together!

I found a similar solved discussion that might help you: ValueError: doc_id not found while using KnowledgeGraphIndex. The issue was related to using storage_context.docstore instead of storage_context.document_store ^[1].

To successfully build, save, and load a KnowledgeGraphIndex, ensuring it retains data and functions correctly, follow these steps:

1. Build the KnowledgeGraphIndex:

   graph_store = SimpleGraphStore()
   storage_context = StorageContext.from_defaults(graph_store=graph_store)
   root_id = "kgindex"
   index = KnowledgeGraphIndex.from_documents(
       documents=documents,
       max_triplets_per_chunk=2,
       storage_context=storage_context,
       embed_model=embed_model,
       include_embeddings=True,
       show_progress=True
   )
   index.set_index_id(root_id)
   storage_context.persist(persist_dir="./index")

2. Load the KnowledgeGraphIndex:

   storage_dir = "./index"
   graph_store = SimpleGraphStore()
   storage_context = StorageContext.from_defaults(graph_store=graph_store)
   storage_context.load(persist_dir=storage_dir)
   index = load_index_from_storage(storage_context, index_id=root_id)
   query_engine = index.as_query_engine(similarity_top_k=2)

3. Convert to networkx graph:

   g = index.get_networkx_graph()
   G = nx.DiGraph(g)

Ensure that the persist_dir is correctly specified and that the index_id matches when loading the index. This should help in retaining the data and functionality of your KnowledgeGraphIndex ^[2][3][4][5].

To continue talking to Dosu, mention @dosu.

[bilalkhan1011]

I have tried this previously numerous times and all of the many iterations of storing and loading which still doesn't work...