Machine Learning Mastery offers developers resources and tutorials on machine learning algorithms, techniques, and applications. Developers can learn about supervised and unsupervised learning methods, deep learning frameworks, and practical machine learning projects. Additionally, the blog covers topics such as data preprocessing, model evaluation, and hyperparameter tuning, providing  insights for both beginners and experienced practitioners in the field of machine learning.

Machine Learning Mastery

Graph RAG is gaining popularity for its ability to organize retrieved data as a graph, connecting documents through nodes and edges to provide comprehensive and insightful responses. This method addresses the limitations of traditional Retrieval-Augmented Generation (RAG) systems, which often fail to connect fragmented information across multiple documents. The post details the step-by-step implementation of Graph RAG using LlamaIndex, including key processes like breaking down documents into text chunks, identifying nodes and edges, summarizing elements, and building communities for more effective data reasoning and responses.

Building a Graph RAG System: A Step-by-Step Approach

Step-by-Step Implementation of GraphRAG with LlamaIndex

<p><a href="https://app.daily.dev/import" data-mention-id="IQVt4AejeUqaHQcdXtczw" data-mention-username="import">@import</a> re
from llama_index.core.graph_stores import SimplePropertyGraphStore
import networkx as nx
from graspologic.partition import hierarchical_leiden</p>
<p>from llama_index.core.llms import ChatMessage
class GraphRAGStore(SimplePropertyGraphStore):
community_summary = {}
max_cluster_size = 5</p>
<pre><code>def generate_community_summary(self, text):
    &quot;&quot;&quot;Generate summary for a given text using an LLM.&quot;&quot;&quot;
    messages = [
        ChatMessage(
            role=&quot;system&quot;,
            content=(
                &quot;You are provided with a set of relationships from a knowledge graph, each represented as &quot;
                &quot;entity1-&gt;entity2-&gt;relation-&gt;relationship_description. Your task is to create a summary of these &quot;
                &quot;relationships. The summary should include the names of the entities involved and a concise synthesis &quot;
                &quot;of the relationship descriptions. The goal is to capture the most critical and relevant details that &quot;
                &quot;highlight the nature and significance of each relationship. Ensure that the summary is coherent and &quot;
                &quot;integrates the information in a way that emphasizes the key aspects of the relationships.&quot;
            ),
        ),
        ChatMessage(role=&quot;user&quot;, content=text),
    ]
    response = OpenAI().chat(messages)
    clean_response = re.sub(r&quot;^assistant:\s*&quot;, &quot;&quot;, str(response)).strip()
    return clean_response

def build_communities(self):
    &quot;&quot;&quot;Builds communities from the graph and summarizes them.&quot;&quot;&quot;
    nx_graph = self._create_nx_graph()
    community_hierarchical_clusters = hierarchical_leiden(
        nx_graph, max_cluster_size=self.max_cluster_size
    )
    community_info = self._collect_community_info(
        nx_graph, community_hierarchical_clusters
    )
    self._summarize_communities(community_info)

def _create_nx_graph(self):
    &quot;&quot;&quot;Converts internal graph representation to NetworkX graph.&quot;&quot;&quot;
    nx_graph = nx.Graph()
    for node in self.graph.nodes.values():
        nx_graph.add_node(str(node))
    for relation in self.graph.relations.values():
        nx_graph.add_edge(
            relation.source_id,
            relation.target_id,
            relationship=relation.label,
            description=relation.properties[&quot;relationship_description&quot;],
        )
    return nx_graph

def _collect_community_info(self, nx_graph, clusters):
    &quot;&quot;&quot;Collect detailed information for each node based on their community.&quot;&quot;&quot;
    community_mapping = {item.node: item.cluster for item in clusters}
    community_info = {}
    for item in clusters:
        cluster_id = item.cluster
        node = item.node
        if cluster_id not in community_info:
            community_info[cluster_id] = []

        for neighbor in nx_graph.neighbors(node):
            if community_mapping[neighbor] == cluster_id:
                edge_data = nx_graph.get_edge_data(node, neighbor)
                if edge_data:
                    detail = f&quot;{node} -&gt; {neighbor} -&gt; {edge_data['relationship']} -&gt; {edge_data['description']}&quot;
                    community_info[cluster_id].append(detail)
    return community_info

def _summarize_communities(self, community_info):
    &quot;&quot;&quot;Generate and store summaries for each community.&quot;&quot;&quot;
    for community_id, details in community_info.items():
        details_text = (
            &quot;\n&quot;.join(details) + &quot;.&quot;
        )  # Ensure it ends with a period
        self.community_summary[
            community_id
        ] = self.generate_community_summary(details_text)

def get_community_summaries(self):
    &quot;&quot;&quot;Returns the community summaries, building them if not already done.&quot;&quot;&quot;
    if not self.community_summary:
        self.build_communities()
    return self.community_summary
</code></pre>
