File size: 3,789 Bytes
a4ab72e | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 | import json
import networkx as nx
import os
from groq import Groq
from dotenv import load_dotenv
load_dotenv()
client = Groq(api_key=os.getenv("GROQ_API_KEY"))
def build_graph() -> nx.Graph:
chunks = json.load(
open(
"data/chunks.json",
encoding="utf-8"
)
)
entities = json.load(
open(
"data/entities.json",
encoding="utf-8"
)
)
G = nx.Graph()
for c in chunks:
G.add_node(
c["chunk_id"],
type="chunk",
text=c["text"],
company=c["company"],
doc=c["doc_name"]
)
G.add_node(
c["company"],
type="company"
)
G.add_node(
c["doc_name"],
type="filing"
)
G.add_edge(
c["company"],
c["doc_name"],
rel="FILED"
)
G.add_edge(
c["doc_name"],
c["chunk_id"],
rel="CONTAINS"
)
for e in entities:
for ent in e["entities"]:
ent_id = (
f"{ent['label']}_"
f"{ent['text'][:50]}"
)
G.add_node(
ent_id,
type="entity",
label=ent["label"],
text=ent["text"]
)
G.add_edge(
e["chunk_id"],
ent_id,
rel="MENTIONS"
)
return G
def query_graphrag(
question: str,
G: nx.Graph
) -> dict:
stopwords = {
"what", "was", "the", "in",
"of", "a", "an", "is",
"are", "how", "did",
"does", "we", "if",
"that", "you", "as",
"based", "on", "which",
"has", "for", "by", "per"
}
keywords = [
w.lower().strip("?.,")
for w in question.split()
if (
w.lower() not in stopwords
and len(w) > 2
)
]
chunk_scores = {}
for node, data in G.nodes(data=True):
if data.get("type") != "chunk":
continue
text = (
data.get("text", "")
+ " "
+ data.get("company", "")
).lower()
score = sum(
1
for kw in keywords
if kw in text
)
if score > 0:
chunk_scores[node] = score
top_chunks = sorted(
chunk_scores,
key=chunk_scores.get,
reverse=True
)[:1]
if not top_chunks:
top_chunks = [
n
for n, d in G.nodes(data=True)
if d.get("type") == "chunk"
][:1]
context = "\n\n".join([
G.nodes[n].get(
"text",
""
)
for n in top_chunks
])
prompt = (
f"Context:\n{context}\n\n"
f"Question: {question}\n"
f"Answer:"
)
input_tokens = len(
prompt.split()
)
response = client.chat.completions.create(
model="llama-3.1-8b-instant",
messages=[{
"role": "user",
"content": prompt
}],
max_tokens=200
)
answer = (
response
.choices[0]
.message.content
)
total_tokens = (
response
.usage
.total_tokens
)
return {
"answer": answer,
"input_tokens": input_tokens,
"total_tokens": total_tokens,
"context_chunks": len(top_chunks)
}
if __name__ == "__main__":
G = build_graph()
print(
f"Graph: "
f"{G.number_of_nodes()} nodes, "
f"{G.number_of_edges()} edges"
)
result = query_graphrag(
"What was Apple's revenue in 2022?",
G
)
print(result) |