vector index: poor recall after many deletions #67
Description
Ladybug version
v0.12.0
What operating system are you using?
macOS 15
What happened?
Issue: if we delete all records from a vector index, this index will be broken and query result will be forever empty.
seemed @1amageek (THANKS!!) tried with a fix kuzudb/kuzu#6046 , while i cherry-picked it to 0.11.1(i am using 0.11.1, confirm lb 0.12.0 with this bug, too) and the issue not fixed
Are there known steps to reproduce?
Reproduction:
# pip install sentence-transformers
import real_ladybug as lb
import os
db = lb.Database("example.lbug")
conn = lb.Connection(db)
from sentence_transformers import SentenceTransformer
model = SentenceTransformer("all-MiniLM-L6-v2")
conn.execute("INSTALL vector; LOAD vector;")
conn.execute("CREATE NODE TABLE Book(id SERIAL PRIMARY KEY, title STRING, title_embedding FLOAT[384], published_year INT64);")
conn.execute("CREATE NODE TABLE Publisher(name STRING PRIMARY KEY);")
conn.execute("CREATE REL TABLE PublishedBy(FROM Book TO Publisher);")
titles = [
"The Quantum World",
"Chronicles of the Universe",
]
publishers = ["Harvard University Press", "Independent Publisher", "Pearson", "McGraw-Hill Ryerson", "O'Reilly"]
published_years = [2004, 2022, 2019, 2010, 2015]
for title, published_year in zip(titles, published_years):
embeddings = model.encode(title).tolist()
conn.execute(
"""
CREATE (b:Book {
title: $title,
title_embedding: $embeddings,
published_year: $year
});""",
{"title": title, "year": published_year, "embeddings": embeddings}
)
print(f"Inserted book: {title}")
conn.execute("""
CALL CREATE_VECTOR_INDEX(
'Book',
'book_title_index',
'title_embedding',
metric := 'l2'
);
""")
query_vector = model.encode("quantum machine learning").tolist()
print("query QUERY_VECTOR_INDEX")
result = conn.execute(
"""
CALL QUERY_VECTOR_INDEX(
'Book',
'book_title_index',
$query_vector,
$limit,
efs := 500
)
RETURN node.title
ORDER BY distance;
""",
{"query_vector": query_vector, "limit": 2})
print("query result:\n")
print(result.get_as_df())
conn.execute("MATCH (b:Book) DETACH DELETE b;")
print("Deleted all Book nodes and their relationships\n\n")
for title, published_year in zip(titles, published_years):
embeddings = model.encode(title).tolist()
conn.execute(
"""
CREATE (b:Book {
title: $title,
title_embedding: $embeddings,
published_year: $year
});""",
{"title": title, "year": published_year, "embeddings": embeddings}
)
print(f"Inserted book: {title}")
print("query QUERY_VECTOR_INDEX")
result = conn.execute(
"""
CALL QUERY_VECTOR_INDEX(
'Book',
'book_title_index',
$query_vector,
$limit,
efs := 500
)
RETURN node.title
ORDER BY distance;
""",
{"query_vector": query_vector, "limit": 2})
print("query result:\n")
print(result.get_as_df()) # here we can see the query result is empty anyway from now on.