[Question]: Knowledge Graph is blank cant't display

[lxljw1314]

Describe your problem

I have deployed Ragflow in a private environment and used the knowledge graph parsing method for analysis. However, after the parsing is completed, the knowledge graph is blank, while the mind map has results.
I tried several embedding models, but the result was the same for all of them; however, the graph displays normally when using the official demo environment provided. please help！

the server log :

2024-12-23 11:22:57,922 INFO 33 172.22.0.6 - - [23/Dec/2024 11:22:57] "GET /v1/document/list?kb_id=7174a526c07911ef9ec30242ac160006&keywords=&page_size=10&page=1 HTTP/1.1" 200 - 2024-12-23 11:22:57,980 INFO 35 From minio(0.014944203197956085) 测试笔录.txt/测试笔录.txt 2024-12-23 11:22:58,003 INFO 35 set_progress(34d74b00c0dd11efa8cb0242ac160006), progress: 0.1, progress_msg: Page(1~100000001): Start to parse. 2024-12-23 11:22:58,026 INFO 35 set_progress(34d74b00c0dd11efa8cb0242ac160006), progress: 0.8, progress_msg: Page(1~100000001): Finish parsing. 2024-12-23 11:22:58,104 INFO 35 set_progress(34d74b00c0dd11efa8cb0242ac160006), progress: 0.5, progress_msg: Page(1~100000001): Extracting entities. 2024-12-23 11:23:10,064 INFO 35 HTTP Request: POST http://10.12.4.31:11434/api/chat "HTTP/1.1 200 OK" 2024-12-23 11:23:25,845 INFO 35 task_consumer_0 reported heartbeat: {"name": "task_consumer_0", "now": "2024-12-23T11:23:25.843712", "boot_at": "2024-12-23T11:13:55.255815", "pending": 1, "lag": 0, "done": 0, "failed": 0, "current": {"id": "34d74b00c0dd11efa8cb0242ac160006", "doc_id": "6c07af94c08211efbc000242ac160006", "from_page": 0, "to_page": 100000000, "retry_count": 0, "kb_id": "7174a526c07911ef9ec30242ac160006", "parser_id": "knowledge_graph", "parser_config": {"entity_types": ["organization", "person", "location", "event", "time"], "chunk_token_num": 8192, "delimiter": "\\n!?;\u3002\uff1b\uff01\uff1f"}, "name": "\u6d4b\u8bd5\u7b14\u5f55.txt", "type": "doc", "location": "\u6d4b\u8bd5\u7b14\u5f55.txt", "size": 3633, "tenant_id": "fed540f0c03411ef87ab0242ac160006", "language": "Chinese", "embd_id": "quentinz/bge-large-zh-v1.5@Ollama", "pagerank": 0, "img2txt_id": "llava@Ollama", "asr_id": "", "llm_id": "qwen2@Ollama", "update_time": 1734924177869}} 2024-12-23 11:23:52,021 INFO 35 HTTP Request: POST http://10.12.4.31:11434/api/chat "HTTP/1.1 200 OK" 2024-12-23 11:23:52,029 INFO 35 set_progress(34d74b00c0dd11efa8cb0242ac160006), progress: None, progress_msg: Page(1~100000001): 1/1, elapsed: 53.93099748343229s, used tokens: 3371 2024-12-23 11:23:52,042 INFO 35 set_progress(34d74b00c0dd11efa8cb0242ac160006), progress: 0.5, progress_msg: Page(1~100000001): Entities extraction progress ... 1/1 2024-12-23 11:23:52,053 INFO 35 set_progress(34d74b00c0dd11efa8cb0242ac160006), progress: 0.6, progress_msg: Page(1~100000001): Extracting community reports. 2024-12-23 11:23:52,068 WARNING 35 /ragflow/.venv/lib/python3.10/site-packages/networkx/readwrite/json_graph/node_link.py:142: FutureWarning: The default value will be edges="edges" in NetworkX 3.6.

To make this warning go away, explicitly set the edges kwarg, e.g.:

nx.node_link_data(G, edges="links") to preserve current behavior, or
nx.node_link_data(G, edges="edges") for forward compatibility.
warnings.warn(

2024-12-23 11:23:52,072 INFO 35 set_progress(34d74b00c0dd11efa8cb0242ac160006), progress: 0.75, progress_msg: Page(1100000001): Extracting mind graph.
2024-12-23 11:23:55,878 INFO 35 task_consumer_0 reported heartbeat: {"name": "task_consumer_0", "now": "2024-12-23T11:23:55.876698", "boot_at": "2024-12-23T11:13:55.255815", "pending": 1, "lag": 0, "done": 0, "failed": 0, "current": {"id": "34d74b00c0dd11efa8cb0242ac160006", "doc_id": "6c07af94c08211efbc000242ac160006", "from_page": 0, "to_page": 100000000, "retry_count": 0, "kb_id": "7174a526c07911ef9ec30242ac160006", "parser_id": "knowledge_graph", "parser_config": {"entity_types": ["organization", "person", "location", "event", "time"], "chunk_token_num": 8192, "delimiter": "\n!?;\u3002\uff1b\uff01\uff1f"}, "name": "\u6d4b\u8bd5\u7b14\u5f55.txt", "type": "doc", "location": "\u6d4b\u8bd5\u7b14\u5f55.txt", "size": 3633, "tenant_id": "fed540f0c03411ef87ab0242ac160006", "language": "Chinese", "embd_id": "quentinz/bge-large-zh-v1.5@Ollama", "pagerank": 0, "img2txt_id": "llava@Ollama", "asr_id": "", "llm_id": "qwen2@Ollama", "update_time": 1734924177869}}
2024-12-23 11:24:25,911 INFO 35 task_consumer_0 reported heartbeat: {"name": "task_consumer_0", "now": "2024-12-23T11:24:25.909271", "boot_at": "2024-12-23T11:13:55.255815", "pending": 1, "lag": 0, "done": 0, "failed": 0, "current": {"id": "34d74b00c0dd11efa8cb0242ac160006", "doc_id": "6c07af94c08211efbc000242ac160006", "from_page": 0, "to_page": 100000000, "retry_count": 0, "kb_id": "7174a526c07911ef9ec30242ac160006", "parser_id": "knowledge_graph", "parser_config": {"entity_types": ["organization", "person", "location", "event", "time"], "chunk_token_num": 8192, "delimiter": "\n!?;\u3002\uff1b\uff01\uff1f"}, "name": "\u6d4b\u8bd5\u7b14\u5f55.txt", "type": "doc", "location": "\u6d4b\u8bd5\u7b14\u5f55.txt", "size": 3633, "tenant_id": "fed540f0c03411ef87ab0242ac160006", "language": "Chinese", "embd_id": "quentinz/bge-large-zh-v1.5@Ollama", "pagerank": 0, "img2txt_id": "llava@Ollama", "asr_id": "", "llm_id": "qwen2@Ollama", "update_time": 1734924177869}}
2024-12-23 11:24:55,944 INFO 35 task_consumer_0 reported heartbeat: {"name": "task_consumer_0", "now": "2024-12-23T11:24:55.942688", "boot_at": "2024-12-23T11:13:55.255815", "pending": 1, "lag": 0, "done": 0, "failed": 0, "current": {"id": "34d74b00c0dd11efa8cb0242ac160006", "doc_id": "6c07af94c08211efbc000242ac160006", "from_page": 0, "to_page": 100000000, "retry_count": 0, "kb_id": "7174a526c07911ef9ec30242ac160006", "parser_id": "knowledge_graph", "parser_config": {"entity_types": ["organization", "person", "location", "event", "time"], "chunk_token_num": 8192, "delimiter": "\n!?;\u3002\uff1b\uff01\uff1f"}, "name": "\u6d4b\u8bd5\u7b14\u5f55.txt", "type": "doc", "location": "\u6d4b\u8bd5\u7b14\u5f55.txt", "size": 3633, "tenant_id": "fed540f0c03411ef87ab0242ac160006", "language": "Chinese", "embd_id": "quentinz/bge-large-zh-v1.5@Ollama", "pagerank": 0, "img2txt_id": "llava@Ollama", "asr_id": "", "llm_id": "qwen2@Ollama", "update_time": 1734924177869}}
2024-12-23 11:25:04,763 INFO 35 HTTP Request: POST http://10.12.4.31:11434/api/chat "HTTP/1.1 200 OK"
2024-12-23 11:25:04,821 INFO 35 Chunking(126.85584738105536) 测试笔录.txt/测试笔录.txt done
2024-12-23 11:25:04,822 INFO 35 MINIO PUT(测试笔录.txt):0
2024-12-23 11:25:04,822 INFO 35 Build document 测试笔录.txt: 126.86s
2024-12-23 11:25:04,835 INFO 35 set_progress(34d74b00c0dd11efa8cb0242ac160006), progress: None, progress_msg: Page(1100000001): Generate 3 chunks
2024-12-23 11:25:06,132 INFO 35 HTTP Request: POST http://10.12.4.31:11434/api/embeddings "HTTP/1.1 200 OK"
2024-12-23 11:25:06,295 INFO 35 HTTP Request: POST http://10.12.4.31:11434/api/embeddings "HTTP/1.1 200 OK"
2024-12-23 11:25:06,332 INFO 35 HTTP Request: POST http://10.12.4.31:11434/api/embeddings "HTTP/1.1 200 OK"
2024-12-23 11:25:06,343 INFO 35 set_progress(34d74b00c0dd11efa8cb0242ac160006), progress: 0.6333333333333333, progress_msg:
2024-12-23 11:25:06,496 INFO 35 HTTP Request: POST http://10.12.4.31:11434/api/embeddings "HTTP/1.1 200 OK"
2024-12-23 11:25:07,006 INFO 35 HTTP Request: POST http://10.12.4.31:11434/api/embeddings "HTTP/1.1 200 OK"
2024-12-23 11:25:07,675 INFO 35 HTTP Request: POST http://10.12.4.31:11434/api/embeddings "HTTP/1.1 200 OK"
2024-12-23 11:25:07,686 INFO 35 set_progress(34d74b00c0dd11efa8cb0242ac160006), progress: 0.7666666666666666, progress_msg:
2024-12-23 11:25:07,689 INFO 35 Embedding chunks (2.85s)
2024-12-23 11:25:07,692 INFO 35 set_progress(34d74b00c0dd11efa8cb0242ac160006), progress: None, progress_msg: Page(1100000001): Embedding chunks (2.85s)
2024-12-23 11:25:07,707 INFO 35 HEAD http://es01:9200/ragflow_fed540f0c03411ef87ab0242ac160006 [status:200 duration:0.011s]
2024-12-23 11:25:07,818 INFO 35 PUT http://es01:9200/ragflow_fed540f0c03411ef87ab0242ac160006/_bulk?refresh=false&timeout=60s [status:200 duration:0.105s]
2024-12-23 11:25:07,823 INFO 35 set_progress(34d74b00c0dd11efa8cb0242ac160006), progress: 0.8333333333333334, progress_msg:
2024-12-23 11:25:07,832 INFO 35 Indexing doc(测试笔录.txt), page(0-100000000), chunks(3), elapsed: 0.12
2024-12-23 11:25:07,842 INFO 35 set_progress(34d74b00c0dd11efa8cb0242ac160006), progress: 1.0, progress_msg: Page(1100000001): Done (0.13s)
2024-12-23 11:25:07,849 INFO 35 Chunk doc(测试笔录.txt), page(0-100000000), chunks(3), token(768), elapsed:0.13
2024-12-23 11:25:07,851 INFO 35 handle_task done for task {"id": "34d74b00c0dd11efa8cb0242ac160006", "doc_id": "6c07af94c08211efbc000242ac160006", "from_page": 0, "to_page": 100000000, "retry_count": 0, "kb_id": "7174a526c07911ef9ec30242ac160006", "parser_id": "knowledge_graph", "parser_config": {"entity_types": ["organization", "person", "location", "event", "time"], "chunk_token_num": 8192, "delimiter": "\n!?;\u3002\uff1b\uff01\uff1f", "layout_recognize": true}, "name": "\u6d4b\u8bd5\u7b14\u5f55.txt", "type": "doc", "location": "\u6d4b\u8bd5\u7b14\u5f55.txt", "size": 3633, "tenant_id": "fed540f0c03411ef87ab0242ac160006", "language": "Chinese", "embd_id": "quentinz/bge-large-zh-v1.5@Ollama", "pagerank": 0, "img2txt_id": "llava@Ollama", "asr_id": "", "llm_id": "qwen2@Ollama", "update_time": 1734924177869}
2024-12-23 11:25:25,975 INFO 35 task_consumer_0 reported heartbeat: {"name": "task_consumer_0", "now": "2024-12-23T11:25:25.974290", "boot_at": "2024-12-23T11:13:55.255815", "pending": 0, "lag": 0, "done": 1, "failed": 0, "current": null}`

[lxljw1314]

and the right result in demo env is like this , Why is the execution result different from what I deployed locally?

[KevinHuSh]

The graph extraction primarily depends on LLM.
So, which LLM did you choose?

[dromeuf]

The graph extraction primarily depends on LLM. So, which LLM did you choose?

Hi Kevin, I have the same problem on last v15.0 with Knowledge Graph. I use snowflake-arctic-embed2 for embedding and qwen2.5:14b for LLM system model settings. The result graph is blank not appears for Know & for Mind. I am testing embedding & small local LLM... result. I have to say that I'm dubious about the results so far.

I'd like to take this opportunity to ask you which LLM RAPTOR depends on for Paper/Book/Gen chunking method (I assume the one defined by default in system Model providers -> System model settings?).

Kind regards and thanks for your great work.

[KevinHuSh]

RAPTOR uses the llm set in system Model providers -> System model settings