[HN Gopher] Show HN: FastGraphRAG - Better RAG using good old Pa...
___________________________________________________________________
Show HN: FastGraphRAG - Better RAG using good old PageRank
Hey there HN! We're Antonio, Luca, and Yuhang, and we're excited to
introduce Fast GraphRAG, an open-source RAG approach that leverages
knowledge graphs and the 25 years old PageRank for better
information retrieval and reasoning. Building a good RAG pipeline
these days takes a lot of manual optimizations. Most engineers
intuitively start from naive RAG: throw everything in a vector
database and hope that semantic search is powerful enough. This can
work for use cases where accuracy isn't too important and
hallucinations are tolerable, but it doesn't work for more
difficult queries that involve multi-hop reasoning or more advanced
domain understanding. Also, it's impossible to debug it. To
address these limitations, many engineers find themselves adding
extra layers like agent-based preprocessing, custom embeddings,
reranking mechanisms, and hybrid search strategies. Much like the
early days of machine learning when we manually crafted feature
vectors to squeeze out marginal gains, building an effective RAG
system often becomes an exercise in crafting engineering "hacks."
Earlier this year, Microsoft seeded the idea of using Knowledge
Graphs for RAG and published GraphRAG - i.e. RAG with Knowledge
Graphs. We believe that there is an incredible potential in this
idea, but existing implementations are naive in the way they create
and explore the graph. That's why we developed Fast GraphRAG with a
new algorithmic approach using good old PageRank. There are two
main challenges when building a reliable RAG system: (1) Data
Noise: Real-world data is often messy. Customer support tickets,
chat logs, and other conversational data can include a lot of
irrelevant information. If you push noisy data into a vector
database, you're likely to get noisy results. (2) Domain
Specialization: For complex use cases, a RAG system must understand
the domain-specific context. This requires creating representations
that capture not just the words but the deeper relationships and
structures within the data. Our solution builds on these insights
by incorporating knowledge graphs into the RAG pipeline. Knowledge
graphs store entities and their relationships, and can help
structure data in a way that enables more accurate and context-
aware information retrieval. 12 years ago Google announced the
knowledge graph we all know about [1]. It was a pioneering move.
Now we have LLMs, meaning that people can finally do RAG on their
own data with tools that can be as powerful as Google's original
idea. Before we built this, Antonio was at Amazon, while Luca and
Yuhang were finishing their PhDs at Oxford. We had been thinking
about this problem for years and we always loved the parallel
between pagerank and the human memory [2]. We believe that
searching for memories is incredibly similar to searching the web.
Here's how it works: - Entity and Relationship Extraction: Fast
GraphRAG uses LLMs to extract entities and their relationships from
your data and stores them in a graph format [3]. - Query
Processing: When you make a query, Fast GraphRAG starts by finding
the most relevant entities using vector search, then runs a
personalized PageRank algorithm to determine the most important
"memories" or pieces of information related to the query [4]. -
Incremental Updates: Unlike other graph-based RAG systems, Fast
GraphRAG natively supports incremental data insertions. This means
you can continuously add new data without reprocessing the entire
graph. - Faster: These design choices make our algorithm faster
and more affordable to run than other graph-based RAG systems
because we eliminate the need for communities and clustering.
Suppose you're analyzing a book and want to focus on character
interactions, locations, and significant events:
from fast_graphrag import GraphRAG DOMAIN = "Analyze
this story and identify the characters. Focus on how they interact
with each other, the locations they explore, and their
relationships." EXAMPLE_QUERIES = [ "What is
the significance of Christmas Eve in A Christmas Carol?",
"How does the setting of Victorian London contribute to the story's
themes?", "Describe the chain of events that leads to
Scrooge's transformation.", "How does Dickens use the
different spirits (Past, Present, and Future) to guide Scrooge?",
"Why does Dickens choose to divide the story into \"staves\" rather
than chapters?" ] ENTITY_TYPES = ["Character",
"Animal", "Place", "Object", "Activity", "Event"] grag
= GraphRAG( working_dir="./book_example",
domain=DOMAIN,
example_queries="\n".join(EXAMPLE_QUERIES),
entity_types=ENTITY_TYPES ) with
open("./book.txt") as f: grag.insert(f.read())
print(grag.query("Who is Scrooge?").response) This code
creates a domain-specific knowledge graph based on your data,
example queries, and specified entity types. Then you can query it
in plain English while it automatically handles all the data
fetching, entity extractions, co-reference resolutions, memory
elections, etc. When you add new data, locking and checkpointing is
handled for you as well. This is the kind of infrastructure that
GenAI apps need to handle large-scale real-world data. Our goal is
to give you this infrastructure so that you can focus on what's
important: building great apps for your users without having to
care about manually engineering a retrieval pipeline. In the
managed service, we also have a suite of UI tools for you to
explore and debug your knowledge graph. We have a free hosted
solution with up to 100 monthly requests. When you're ready to
grow, we have paid plans that scale with you. And of course you can
self host our open-source engine. Give us a spin today at
https://circlemind.co and see our code at
https://github.com/circlemind-ai/fast-graphrag We'd love feedback
:) [1] https://blog.google/products/search/introducing-knowledge-
gr... [2] Griffiths, T. L., Steyvers, M., & Firl, A. (2007).
Google and the Mind: Predicting Fluency with PageRank.
Psychological Science, 18(12), 1069-1076.
http://www.jstor.org/stable/40064705 [3] Similarly to Microsoft's
GraphRAG: https://github.com/microsoft/graphrag [4] Similarly to
OSU's HippoRAG: https://github.com/OSU-NLP-Group/HippoRAG
https://vhs.charm.sh/vhs-4fCicgsbsc7UX0pemOcsMp.gif
Author : liukidar
Score : 160 points
Date : 2024-11-18 17:43 UTC (5 hours ago)
(HTM) web link (github.com)
(TXT) w3m dump (github.com)
| dartos wrote:
| Please tell me I'm not the only one that sees the irony in AI
| relying on classic search.
|
| Obviously LLMs are good at some semantic understanding of the
| prompt context and are useful, but the irony is hilarious
| antves wrote:
| I think the main bit here is that the knowledge graph is
| entirely constructed by LLMs. It's not just using a pre-
| existing knowledge graph. It's creating a knowledge graph on
| the fly based on your data.
|
| Navigating the graph, on the other hand, is the perfect task
| for PageRank.
| liukidar wrote:
| Exactly! Also PageRank is used to navigate the graph and find
| "missing links" between the concepts selected from the query
| using semantic search via LLMs (so to be able to find
| information to answer questions that require multi-hop or
| complex reasoning in one go).
| dartos wrote:
| Makes perfect sense.
|
| The semantic understanding capabilities fit well for creating
| knowledge graphs.
| IanCal wrote:
| I don't get what the irony is here.
| chefandy wrote:
| Not who you're replying to, but from my vantage point,
| marketing folks seem to be pushing LLM products as
| replacements for traditional search products. I think what
| the post is proposing makes perfect sense from a technical
| perspective, though. The utility of LLMs will come down to
| good old-fashioned product design, leveraging existing
| concepts, and novel technical innovation rather than just
| dumping quintillions of dollars into increasingly large
| models and hardware until it does everything for us.
| dartos wrote:
| Exactly this.
|
| I work in the LLM-augmented search space, so I might be a
| little too tuned in on this subject.
| adultSwim wrote:
| Can this be used with LLMs other than the OpenAI API?
| antves wrote:
| Yes, it works out-of-the-box with any OpenAI compatible API,
| including Ollama.
|
| You can check out our example at https://github.com/circlemind-
| ai/fast-graphrag/blob/main/exa...
| youngNed wrote:
| it would be nice to see an example that uses ollama - given
| that ollamas embeddings endpoint is a bit... different, I
| can't quite figure this out
| liukidar wrote:
| Hey! Our todo list is a bit swamped with things right now,
| but we'll try to have a look at that as soon as possible.
| On the Ollama github I found contrasting information:
| https://github.com/ollama/ollama/issues/2416 and
| https://github.com/ollama/ollama/pull/2925 They also
| suggest to look at this:
| https://github.com/severian42/GraphRAG-Local-
| UI/blob/main/em...
|
| Hope this can help!
| deepsquirrelnet wrote:
| This is cool! How is the graph stored and queried? I'm familiar
| with graph databases, but I don't see that as a dependency.
|
| Have you tried the sciphi triplex model for extraction? I've
| tried to do some extraction before, but got inconsistent results
| if I extracted the chunks multiple times consecutively.
| liukidar wrote:
| The graph is currently stored using python-igraph. The codebase
| is designed such that it is easy to integrate any graphdb by
| writing a light wrapper around it (we will provide support to
| stuff like neo4j in the near future). We haven't tried triplex
| since we saw that gpt4o-mini is fast and precise enough for now
| (and we use it not only for extraction of entities and
| relationships, but also to get descriptions and resolve
| conflicts), but for sure with fine tuning results should
| improve. The graph is queried by finding an initial set of
| nodes that are relevant to a given query and then running
| personalized pageranking from those nodes to find other
| relevant passages. Currently, we select the inital nodes with
| semantic search both on the whole query and entities extracted
| from it, but we are planning for other exciting additions to
| this method :)
| stephantul wrote:
| Just out of interest: why is every python file prefixed with an
| underscore? I've never seen it before. Is it to avoid collisions
| with package imports? e.g. "types"
| liukidar wrote:
| It is to mark the package as private (in the sense that for
| normal usage you shouldn't need it). We are still writing the
| documentation on how to customize every little bit of the graph
| construction and querying pipeline, once that is ready we will
| expose the right tools (and files) for all of that :) For now
| just go with `from fast_graphrag import GraphRAG` and you
| should be good to go :)
| dvaun wrote:
| It's the standard practice of noting internal/implementation
| details. Users should stick with the public api exported in
| __init__.
| stephantul wrote:
| Interesting, thanks! I've seen it for importable stuff but
| never for modules.
|
| I guess I'm getting old
| fareesh wrote:
| What solutions are folks using to solve queries like "How many of
| these 1000 podcast transcripts have a positive view of Hillary
| Clinton"? Seems like you would need a way to map reduce and
| count? And some kind of agent assigner/router on top of it?
| chefandy wrote:
| I don't have a dev answer, but in case its relevant, I've seen
| commercial services that I imagine are doing something similar
| on the back end-- ground news is one of them. I wish they had
| monthly subs for their top tier plan rather than only annual,
| but it seems like a cool product. I haven't actually used it
| though.
| aspenmayer wrote:
| What feature(s) of the top tier plan do you wish you had? I
| have no idea how their subs work but have seen a few ads for
| the product so have a vague idea that it rates news for bias
| but don't see how that would involve many different tiers of
| subs.
| nkristoffersen wrote:
| We do a lot of things with podcast and other audio media at
| https://listenalert.com
|
| But in general we found the best course of action is simply
| label everything. Because our customers will want those answers
| and rag won't really work at the scale of "all podcasts the
| last 6 months. What is the trend of sentiment Hillary Clinton
| and what about the top topics and entities mentioned nearby".
| So we take a more "brute force" approach :-)
| treprinum wrote:
| Anticipate what kind of questions user might ask, pre-compute
| the answers and store them as natural sentences in a vector
| database.
| avereveard wrote:
| llm can write graph queries
| antves wrote:
| At the moment this repo is designed to handle more RAG-oriented
| use cases, i.e. that require to recall the "top pieces of
| information" relevant to a given question/context. In your
| specific example, right now, FastGraphRAG would select the
| nodes that represent podcasts that are connected to Hilary
| Clinton, feed them to an LLM which would then select the ones
| that are positively associated with her. As a next step, we
| plan to weight the connections between nodes given the query.
| This way, PageRank will explore only edges which carry the
| concept "positively associated with", and only the right
| podcasts would be selected and returned, without having to ask
| an LLM to classify them. Note that this is basically a fuzzy
| join and so it will produce only a "best-effort" answer rather
| than an exact one.
| vasa_ wrote:
| Neat, we are doing something similar with cognee, but are letting
| users define graph ingestion, generation, and retrieval
| themselves instead of making assumptions:
| https://github.com/topoteretes/cognee
| handfuloflight wrote:
| Could this be used for context retrieval and generative
| understanding of codebases?
| antves wrote:
| Yes, feel free to try it out! You can specialize the graph for
| the codebase use-case by configuring the graph prompt, entity
| types, and example questions accordingly.
| zwaps wrote:
| What is the difference to HippoRAG, which seems to be the same
| approach but came our earlier?
| antves wrote:
| HippoRAG is an amazing work and it was a source of inspiration
| for us as noted in the references. There are a couple of
| differences:
|
| (1) FastGraphRAG allows the user to make the graph construction
| opinionated and specialized on a given domain and for a given
| use-case; this allows to clear out all the noise in the data
| and yields better results; (2) Unlike HippoRAG, FastGraphRAG
| initializes PageRank with a mixture of semantic retrieval and
| entity extractions; (3) HippoRAG is the outcome of an academic
| paper, and we saw the need for a more robust and production-
| ready implementation. Our repo is fully typed, includes tests,
| handles retries with Instructor, uses structured outputs, and
| so on.
|
| Moving forward, we see our implementation diverge from HippoRAG
| more radically as we start to introduce new mechanisms such as
| weighted edges and negative PageRank to model repulsors.
| adamgordonbell wrote:
| So what is the answer to "Who is Scrooge?" and is it different /
| better than another approach?
|
| ( Like whole thing in contenxt window for instance? )
|
| Is this approach just for cost savings or does it help get better
| answers and how so?
|
| Could you share a specific example?
| antves wrote:
| Generally speaking RAG comes in the game when it is impractical
| to use large context windows for three reasons: (1) accuracy
| drops as you stuff the context windows, (2) currently, context
| windows do not scale past 1M tokens, and (3) even with caching,
| moving millions of tokens is wasteful and not viable both in
| terms of costs and latency.
|
| So we should really compare this to other RAG approaches. If we
| compare it to vector databases RAG, knowledge graphs have the
| advantage that they model the connections between datapoints.
| This is super important when asking questions that requires to
| reason across multiple pieces of information, i.e. multi-hop
| reasoning.
|
| Also, the graph construction is essentially an exercise in
| cleaning data to extract the knowledge. Let me give you a
| practical example. Let's pretend we're indexing customer
| tickets for creating an AI assistant. If we were to store the
| data on the tickets as it is, we would overwhelm the vector
| database with all the noise coming from the conversational
| nature of this data. With knowledge graphs, we extract only the
| relevant entities and relationships and store the distilled
| knowledge in our graph. At query time, we find the answer over
| a structured data model that contains only clean information
| adamgordonbell wrote:
| Makes sense, but so can you compare it to to RAG then and
| show how an answer is superior and what the context contains
| that makes it superior?
|
| Or how it is close to large context quality of answer with
| lower cost on some specific examples.
|
| It's helpful when a readme contains a demonstration or as I
| said above, a specific example.
| AIorNot wrote:
| This is very cool, I signed up and uploaded a few docs (PDFs) to
| the dashboard
|
| Our Use case: We have been looking at farming out this work
| (analyzing complaince documents (manufacturing paperwork) for our
| AI Startup however we need to understand the potential scale this
| can operate under and the cost model for it to be useful to us
|
| We will have about 300K PDF documents per client and expect about
| a 10% change in that document set, month to month -any GraphRag
| system has to handle documents at scale - we can use S3 as an
| igestion mechanism but have to understand the cost and processing
| time needed for the system to be ready to use duiring:
|
| 1. inital loading 2. regular updates -how do we delete data from
| system for example
|
| cool framework btw..
| antves wrote:
| Thanks! It sounds like we should be able to help. I'd love to
| chat more in detail, feel free to send me a note at antonio
| [at] circlemind.co.
| barlimp wrote:
| Can this method return references to source documents?
| antves wrote:
| Yes. We already support this feature in our managed service
| (https://docs.circlemind.co/essentials/query#include-
| referenc...) and we'll include it in the next open-source
| release too!
| larodi wrote:
| Wonder why this all - here on HN - is not part of the readme .md
| which says absolutely nothing about how and why this all would
| work.
|
| The whole approach to representing the work, including the
| writing here, screams marketing, and the paid offering is the
| only thing made absolutely clear about it.
|
| p.s. I absolutely understand why a knowledge graph is essential
| and THE right approach for RAG, and particularly when vector DBS
| on their own are subpar. But so do know many others and from the
| way the repo is presented it absolutely gives no clue why yours
| is _something_ in respect to other/common-sense graph-RAG-
| somethings.
|
| You see, there are hundreds of smart people out there who can
| easily come to conclusion data needs to be presented as knowledge
| in graph-ontological way and then feed the context with only the
| relevant subgraph. Like, you could've said so much rather than
| asking .0084 cents or whatever for APIs as the headline of a
| presumably open repo.
| azinman2 wrote:
| HN is "for" startups. This is a startup. What's the problem?
| antves wrote:
| I completely agree that the README could do a better job
| explaining the implementation details and our reasoning behind
| key design choices. For instance, we should elaborate on why we
| believe using PageRank offers a more effective exploration
| strategy compared to other GraphRAG approaches.
|
| FastGraphRAG is entirely free to use, even for commercial
| applications, and we're happy to make it accessible to
| everyone. The managed service is how we sustain our business.
| peppertree wrote:
| How does domain and example queries help construct the knowledge
| graph, or is that just context for executing queries.
| antves wrote:
| These are knobs that you can tune to make the graph
| construction more/less opinionated. Generally speaking, the
| more we make it opinionated the better it fits the task.
|
| At a high-level:
|
| (1) Domain: allows you to "talk to the graph constructor". If
| you care particularly about one aspect of your data, this is
| the place to say it. For reference, take a look at some of the
| example prompts on our website (https://circlemind.co/)
|
| (2) Example Queries: if you know what class of questions users
| will ask, it'd be useful to give the system this information so
| that it will "keep these questions in mind" when designing the
| graph. If you don't know which kinds of questions, you can just
| put a couple of high-level questions that you think apply to
| your data.
|
| (3) Entity Types: this has a very high impact on the final
| quality of the graph. Think of these as the types of entities
| that you want to extract from your data, e.g. person, place,
| event, etc
|
| All of the above help construct the knowledge graph so that it
| is specifically designed for your use-case.
| davidsainez wrote:
| Cool! But I'm confused on your pricing. The github page says
| first 100 requests are free but the landing page says to self
| host if you want to use for free. I signed up and used the
| dashboard but I don't see a billing section or option to upgrade
| the account.
| antves wrote:
| Thanks for trying it out! There are two options to use
| FastGraphRAG for free:
|
| (1) Self-hosting our open-source package (2) Using the free
| tier of the managed service, which includes 100 requests
|
| If you wish to upgrade your plan, you can reach out to us at
| support [at] circlemind.co
| bionhoward wrote:
| Since when does "good old PageRank" demand an OpenAI API key?
|
| "You may not: use Output to develop models that compete with
| OpenAI" => they're gonna learn from you and you can't learn from
| them.
|
| Glad we're all so cool with longterm economic downfall of natural
| humans. Our grandkids might not be so glad about it!
| anotherpaulg wrote:
| Super interesting, thanks for sharing. How large a corpus of
| domain specific text do you need to obtain a useful knowledge
| graph?
|
| Aider has been doing PageRank on the call graph of code repos
| since forever. All non trivial code has lots of graph structure
| to support PageRank. So it works really well to find the most
| relevant context in the project related to the currently active
| task.
|
| https://aider.chat/docs/repomap.html#optimizing-the-map
| liukidar wrote:
| We have tried from small novels to full documentations of some
| milion tokens and both seem to create interesting graphs, it
| would be great to hear some feedback as more people start using
| it :)
| LASR wrote:
| So I've done a ton of work in this area.
|
| Few learnings I've collected:
|
| 1. Lexical search with BM25 alone gives you very relevant results
| if you can do some work during ingestion time with an LLM.
|
| 2. Embeddings work well only when the size of the query is
| roughly on the same order of what you're actually storing in the
| embedding store.
|
| 3. Hypothetical answer generation from a query using an LLM, and
| then using that hypothetical answer to query for embeddings works
| really well.
|
| So combining all 3 learnings, we landed on a knowledge
| decomposition and extraction step very similar to yours. But we
| stick a metaprompter to essentially auto-generate the domain /
| entity types.
|
| LLMs are naively bad at identifying the correct level of
| granularity for the decomposed knowledge. One trick we found is
| to ask the LLM to output a mermaid.js mindmap to hierarchically
| break down the input into a tree. At the end of that output, ask
| the LLM to state which level is the appropriate root for a
| knowledge node.
|
| Then the node is used to generate questions that could be
| answered from the knowledge contained in this node. We then index
| the text of these questions and also embed them.
|
| You can directly match the user's query from these questions
| using purely BM25 and get good outputs. But a hybrid approach
| works even better, though not by that much.
|
| Not using LLMs are query time also means we can hierarchically
| walk down the root into deeper and deeper nodes, using the
| embedding similiarity as a cost function for the traversal.
| michelpp wrote:
| PageRank is one of several interesting centrality metrics that
| could be applied to a graph to influence RAG on structural data,
| another one is Triangle Centrality which counts triangles around
| nodes to figure out their centrality based on the concept that
| triangles close relationships into a strong bond, where open
| bonds dilute centrality by drawing weight away from the center:
|
| https://arxiv.org/abs/2105.00110
|
| The paper shows high efficiency compared to other centralities
| like PageRank, however in some research using the GraphBLAS I and
| my coauthors found that TC was slower on a variety of sparse
| graphs than our sparse formulation of PR for graphs up to 1.8
| billion edges, but that TC appears to scale better as graphs get
| larger and is likely more efficient in the trillion edge realm.
|
| https://fossies.org/linux/SuiteSparse/GraphBLAS/Doc/The_Grap...
___________________________________________________________________
(page generated 2024-11-18 23:00 UTC)