Skip to content

Commit 1aed611

Browse files
prateekchhikaraprateekchhikara
authored
Added graph memory (#2403)
1 parent 6c2b131 commit 1aed611

File tree

4 files changed

+299
-0
lines changed

4 files changed

+299
-0
lines changed

‎docs/docs.json

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -56,6 +56,7 @@
5656
"features/async-client",
5757
"features/memory-export",
5858
"features/webhooks",
59+
"features/graph-memory",
5960
"features/feedback-mechanism"
6061
]
6162
}

‎docs/features/graph-memory.mdx

Lines changed: 295 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,295 @@
1+
---
2+
title: Graph Memory
3+
icon: "circle-nodes"
4+
iconType: "solid"
5+
description: "Enable graph-based memory retrieval for more contextually relevant results"
6+
---
7+
8+
## Overview
9+
10+
Graph Memory enhances memory pipeline by creating relationships between entities in your data. It builds a network of interconnected information for more contextually relevant search results.
11+
12+
This feature allows your AI applications to understand connections between entities, providing richer context for responses. It's ideal for applications needing relationship tracking and nuanced information retrieval across related memories.
13+
14+
## How Graph Memory Works
15+
16+
The Graph Memory feature analyzes how each entity connects and relates to each other. When enabled:
17+
18+
1. Mem0 automatically builds a graph representation of entities
19+
2. Retrieval considers graph relationships between entities
20+
3. Results include entities that may be contextually important even if they're not direct semantic matches
21+
22+
## Using Graph Memory
23+
24+
To use Graph Memory, you need to enable it in your API calls by setting the `enable_graph=True` parameter. You'll also need to specify `output_format="v1.1"` to receive the enriched response format.
25+
26+
### Adding Memories with Graph Memory
27+
28+
When adding new memories, enable Graph Memory to automatically build relationships with existing memories:
29+
30+
<CodeGroup>
31+
32+
```python Python
33+
from mem0 import MemoryClient
34+
35+
client = MemoryClient(
36+
api_key="your-api-key",
37+
org_id="your-org-id",
38+
project_id="your-project-id"
39+
)
40+
41+
messages = [
42+
{"role": "user", "content": "My name is Joseph"},
43+
{"role": "assistant", "content": "Hello Joseph, it's nice to meet you!"},
44+
{"role": "user", "content": "I'm from Seattle and I work as a software engineer"}
45+
]
46+
47+
# Enable graph memory when adding
48+
client.add(
49+
messages,
50+
user_id="joseph",
51+
version="v1",
52+
enable_graph=True,
53+
output_format="v1.1"
54+
)
55+
```
56+
57+
```javascript JavaScript
58+
import { MemoryClient } from "mem0";
59+
60+
const client = new MemoryClient({
61+
apiKey: "your-api-key",
62+
orgId: "your-org-id",
63+
projectId: "your-project-id"
64+
});
65+
66+
const messages = [
67+
{ role: "user", content: "My name is Joseph" },
68+
{ role: "assistant", content: "Hello Joseph, it's nice to meet you!" },
69+
{ role: "user", content: "I'm from Seattle and I work as a software engineer" }
70+
];
71+
72+
// Enable graph memory when adding
73+
await client.add({
74+
messages,
75+
userId: "joseph",
76+
version: "v1",
77+
enableGraph: true,
78+
outputFormat: "v1.1"
79+
});
80+
```
81+
82+
```json Output
83+
{
84+
"results": [
85+
{
86+
"memory": "Name is Joseph",
87+
"event": "ADD",
88+
"id": "4a5a417a-fa10-43b5-8c53-a77c45e80438"
89+
},
90+
{
91+
"memory": "Is from Seattle",
92+
"event": "ADD",
93+
"id": "8d268d0f-5452-4714-b27d-ae46f676a49d"
94+
},
95+
{
96+
"memory": "Is a software engineer",
97+
"event": "ADD",
98+
"id": "5f0a184e-ddea-4fe6-9b92-692d6a901df8"
99+
}
100+
]
101+
}
102+
```
103+
</CodeGroup>
104+
105+
The graph memory would look like this:
106+
107+
<Frame>
108+
<img src="/images/graph-platform.png" alt="Graph Memory Visualization showing relationships between entities" />
109+
</Frame>
110+
111+
<Caption>Graph Memory creates a network of relationships between entities, enabling more contextual retrieval</Caption>
112+
113+
114+
<Note>
115+
Response for the graph memory's `add` operation will not be available directly in the response.
116+
As adding graph memories is an asynchronous operation due to heavy processing,
117+
you can use the `get_all()` endpoint to retrieve the memory with the graph metadata.
118+
</Note>
119+
120+
121+
### Searching with Graph Memory
122+
123+
When searching memories, Graph Memory helps retrieve entities that are contextually important even if they're not direct semantic matches.
124+
125+
<CodeGroup>
126+
127+
```python Python
128+
# Search with graph memory enabled
129+
results = client.search(
130+
"what is my name?",
131+
user_id="joseph",
132+
enable_graph=True,
133+
output_format="v1.1"
134+
)
135+
136+
print(results)
137+
```
138+
139+
```javascript JavaScript
140+
// Search with graph memory enabled
141+
const results = await client.search({
142+
query: "what is my name?",
143+
userId: "joseph",
144+
enableGraph: true,
145+
outputFormat: "v1.1"
146+
});
147+
148+
console.log(results);
149+
```
150+
151+
```json Output
152+
{
153+
"results": [
154+
{
155+
"id": "4a5a417a-fa10-43b5-8c53-a77c45e80438",
156+
"memory": "Name is Joseph",
157+
"user_id": "joseph",
158+
"metadata": null,
159+
"categories": ["personal_details"],
160+
"immutable": false,
161+
"created_at": "2025-03-19T09:09:00.146390-07:00",
162+
"updated_at": "2025-03-19T09:09:00.146404-07:00",
163+
"score": 0.3621795393335552
164+
},
165+
{
166+
"id": "8d268d0f-5452-4714-b27d-ae46f676a49d",
167+
"memory": "Is from Seattle",
168+
"user_id": "joseph",
169+
"metadata": null,
170+
"categories": ["personal_details"],
171+
"immutable": false,
172+
"created_at": "2025-03-19T09:09:00.170680-07:00",
173+
"updated_at": "2025-03-19T09:09:00.170692-07:00",
174+
"score": 0.31212713194651254
175+
}
176+
],
177+
"relations": [
178+
{
179+
"source": "joseph",
180+
"source_type": "person",
181+
"relationship": "name",
182+
"target": "joseph",
183+
"target_type": "person",
184+
"score": 0.39
185+
}
186+
]
187+
}
188+
```
189+
190+
</CodeGroup>
191+
192+
### Retrieving All Memories with Graph Memory
193+
194+
When retrieving all memories, Graph Memory provides additional relationship context:
195+
196+
<CodeGroup>
197+
198+
```python Python
199+
# Get all memories with graph context
200+
memories = client.get_all(
201+
user_id="joseph",
202+
enable_graph=True,
203+
output_format="v1.1"
204+
)
205+
206+
print(memories)
207+
```
208+
209+
```javascript JavaScript
210+
// Get all memories with graph context
211+
const memories = await client.getAll({
212+
userId: "joseph",
213+
enableGraph: true,
214+
outputFormat: "v1.1"
215+
});
216+
217+
console.log(memories);
218+
```
219+
220+
```json Output
221+
{
222+
"results": [
223+
{
224+
"id": "5f0a184e-ddea-4fe6-9b92-692d6a901df8",
225+
"memory": "Is a software engineer",
226+
"user_id": "joseph",
227+
"metadata": null,
228+
"categories": ["professional_details"],
229+
"immutable": false,
230+
"created_at": "2025-03-19T09:09:00.194116-07:00",
231+
"updated_at": "2025-03-19T09:09:00.194128-07:00",
232+
},
233+
{
234+
"id": "8d268d0f-5452-4714-b27d-ae46f676a49d",
235+
"memory": "Is from Seattle",
236+
"user_id": "joseph",
237+
"metadata": null,
238+
"categories": ["personal_details"],
239+
"immutable": false,
240+
"created_at": "2025-03-19T09:09:00.170680-07:00",
241+
"updated_at": "2025-03-19T09:09:00.170692-07:00",
242+
},
243+
{
244+
"id": "4a5a417a-fa10-43b5-8c53-a77c45e80438",
245+
"memory": "Name is Joseph",
246+
"user_id": "joseph",
247+
"metadata": null,
248+
"categories": ["personal_details"],
249+
"immutable": false,
250+
"created_at": "2025-03-19T09:09:00.146390-07:00",
251+
"updated_at": "2025-03-19T09:09:00.146404-07:00",
252+
}
253+
],
254+
"relations": [
255+
{
256+
"source": "joseph",
257+
"source_type": "person",
258+
"relationship": "name",
259+
"target": "joseph",
260+
"target_type": "person"
261+
},
262+
{
263+
"source": "joseph",
264+
"source_type": "person",
265+
"relationship": "city",
266+
"target": "seattle",
267+
"target_type": "city"
268+
},
269+
{
270+
"source": "joseph",
271+
"source_type": "person",
272+
"relationship": "job",
273+
"target": "software engineer",
274+
"target_type": "job"
275+
}
276+
]
277+
}
278+
```
279+
280+
</CodeGroup>
281+
282+
## Best Practices
283+
284+
- Enable Graph Memory for applications where understanding context and relationships between memories is important
285+
- Graph Memory works best with a rich history of related conversations
286+
- Consider Graph Memory for long-running assistants that need to track evolving information
287+
288+
## Performance Considerations
289+
290+
Graph Memory requires additional processing and may increase response times slightly for very large memory stores. However, for most use cases, the improved retrieval quality outweighs the minimal performance impact.
291+
292+
If you have any questions, please feel free to reach out to us using one of the following methods:
293+
294+
<Snippet file="get-help.mdx" />
295+

‎docs/features/platform-overview.mdx

Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -36,6 +36,9 @@ Learn about the key features and capabilities that make Mem0 a powerful platform
3636
<Card title="Memory Export" icon="file-export" href="/features/memory-export">
3737
Export memories in structured formats using customizable Pydantic schemas.
3838
</Card>
39+
<Card title="Graph Memory" icon="graph" href="/features/graph-memory">
40+
Add memories in the form of nodes and edges in a graph database and search for related memories.
41+
</Card>
3942
</CardGroup>
4043

4144
## Getting Help

‎docs/images/graph-platform.png

72.2 KB
Loading

0 commit comments

Comments
 (0)