Skip to content

feat: Neo4j vector index method support #1111

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 4 commits into from
Oct 10, 2025
Merged

feat: Neo4j vector index method support #1111

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
0da98f9
feat: Neo4j vector index method support
belloibrahv Oct 2, 2025
c37b57c
feat: integrate Ollama for LLM tasks to enable on-premises testing
belloibrahv Oct 4, 2025
da71d00
Simplify Neo4j vector index
belloibrahv Oct 8, 2025
6ba1167
Fix: Refactor LLM usage and update Neo4j vector index
belloibrahv Oct 9, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 16 additions & 4 deletions examples/docs_to_knowledge_graph/main.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,9 +82,15 @@ def docs_to_kg_flow(
cocoindex.functions.ExtractByLlm(
llm_spec=cocoindex.LlmSpec(
# Supported LLM: https://cocoindex.io/docs/ai/llm
api_type=cocoindex.LlmApiType.OPENAI,
model="gpt-4o",
api_type=cocoindex.LlmApiType.OLLAMA,
model="llama3.2",
address="http://localhost:11434",
),
# Alternative: Use OpenAI API model instead of Ollama
# llm_spec=cocoindex.LlmSpec(
# api_type=cocoindex.LlmApiType.OPENAI,
# model="gpt-4o",
# ),
output_type=DocumentSummary,
instruction="Please summarize the content of the document.",
)
Expand All @@ -100,9 +106,15 @@ def docs_to_kg_flow(
cocoindex.functions.ExtractByLlm(
llm_spec=cocoindex.LlmSpec(
# Supported LLM: https://cocoindex.io/docs/ai/llm
api_type=cocoindex.LlmApiType.OPENAI,
model="gpt-4o",
api_type=cocoindex.LlmApiType.OLLAMA,
model="llama3.2",
address="http://localhost:11434",
),
# Alternative: Use OpenAI API model instead of Ollama
# llm_spec=cocoindex.LlmSpec(
# api_type=cocoindex.LlmApiType.OPENAI,
# model="gpt-4o",
# ),
output_type=list[Relationship],
instruction=(
"Please extract relationships from CocoIndex documents. "
Expand Down
2 changes: 1 addition & 1 deletion src/base/spec.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -384,7 +384,7 @@ impl fmt::Display for VectorSimilarityMetric {
}
}

#[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]
#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
#[serde(tag = "kind")]
pub enum VectorIndexMethod {
Hnsw {
Expand Down
38 changes: 33 additions & 5 deletions src/ops/targets/neo4j.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -568,9 +568,6 @@ impl SetupState {
.map(|f| (f.name.as_str(), &f.value_type.typ))
.collect::<HashMap<_, _>>();
for index_def in index_options.vector_indexes.iter() {
if index_def.method.is_some() {
api_bail!("Vector index method is not configurable for Neo4j yet");
}
sub_components.push(ComponentState {
object_label: schema.elem_type.clone(),
index_def: IndexDef::from_vector_index_def(
Expand All @@ -583,6 +580,7 @@ impl SetupState {
index_def.field_name
)
})?,
index_def.method.clone(),
)?,
});
}
Expand Down Expand Up @@ -644,14 +642,20 @@ enum IndexDef {
field_name: String,
metric: spec::VectorSimilarityMetric,
vector_size: usize,
method: Option<spec::VectorIndexMethod>,
},
}

impl IndexDef {
fn from_vector_index_def(
index_def: &spec::VectorIndexDef,
field_typ: &schema::ValueType,
method: Option<spec::VectorIndexMethod>,
) -> Result<Self> {
let method = index_def.method.clone();
if let Some(spec::VectorIndexMethod::IvfFlat { .. }) = method {
api_bail!("IVFFlat vector index method is not supported for Neo4j");
}
Ok(Self::VectorIndex {
field_name: index_def.field_name.clone(),
vector_size: (match field_typ {
Expand All @@ -664,6 +668,7 @@ impl IndexDef {
api_error!("Vector index field must be a vector with fixed dimension")
})?,
metric: index_def.metric,
method,
})
}
}
Expand Down Expand Up @@ -723,9 +728,14 @@ impl components::SetupOperator for SetupComponentOperator {
field_name,
metric,
vector_size,
method,
} => {
let method_str = method
.as_ref()
.map(|m| format!(", method: {}", m))
.unwrap_or_default();
format!(
"{key_desc} ON {label} (field_name: {field_name}, vector_size: {vector_size}, metric: {metric})",
"{key_desc} ON {label} (field_name: {field_name}, vector_size: {vector_size}, metric: {metric}{method_str})",
)
}
}
Expand All @@ -752,14 +762,32 @@ impl components::SetupOperator for SetupComponentOperator {
field_name,
metric,
vector_size,
method,
} => {
let method_config =
if let Some(spec::VectorIndexMethod::Hnsw { m, ef_construction }) = method {
let mut parts = vec![];
if let Some(m_val) = m {
parts.push(format!("`hnsw.m`: {}", m_val));
}
if let Some(ef_val) = ef_construction {
parts.push(format!("`hnsw.ef_construction`: {}", ef_val));
}
if parts.is_empty() {
"".to_string()
} else {
format!(", {}", parts.join(", "))
}
Copy link
Member

@georgeh0 georgeh0 Oct 5, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This logic is a little bit fragile. A slightly simpler way is to put all index config fields into parts (i.e. for existing ones like "vector.dimensions: {vector_size}" also put into parts. Then we only need a simple parts.join(", ") at last, which will make our logic simpler and more clear.

Copy link
Contributor Author

@belloibrahv belloibrahv Oct 8, 2025 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@georgeh0 Refactored vector config logic using parts.join as requested feedback. kindly help review.

} else {
"".to_string()
};
formatdoc! {"
CREATE VECTOR INDEX {name} IF NOT EXISTS
FOR {matcher} ON {qualifier}.{field_name}
OPTIONS {{
indexConfig: {{
`vector.dimensions`: {vector_size},
`vector.similarity_function`: '{metric}'
`vector.similarity_function`: '{metric}'{method_config}
}}
}}",
name = key.name,
Expand Down