my-project-public

repository

loading code, commits, and activity

repositories

loading repo index

my-project-public — gitlawb

#1	---
#2	title: Graph Memory
#3	description: "Enable graph-based memory retrieval for more contextually relevant results"
#4	---
#5
#6	## Overview
#7
#8	Graph Memory enhances the memory pipeline by creating relationships between entities in your data. It builds a network of interconnected information for more contextually relevant search results.
#9
#10	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.
#11
#12	## How Graph Memory Works
#13
#14	The Graph Memory feature analyzes how each entity connects and relates to each other. When enabled:
#15
#16	1. Mem0 automatically builds a graph representation of entities
#17	2. Vector search returns the top semantic matches (with any reranker you configure)
#18	3. Graph relations are returned alongside those results to provide additional context—they do not reorder the vector hits
#19
#20	## Using Graph Memory
#21
#22	To use Graph Memory, you need to enable it in your API calls by setting the `enable_graph=True` parameter.
#23
#24	### Adding Memories with Graph Memory
#25
#26	When adding new memories, enable Graph Memory to automatically build relationships with existing memories:
#27
#28	<CodeGroup>
#29
#30	```python Python
#31	from mem0 import MemoryClient
#32
#33	client = MemoryClient(
#34	api_key="your-api-key",
#35	org_id="your-org-id",
#36	project_id="your-project-id"
#37	)
#38
#39	messages = [
#40	{"role": "user", "content": "My name is Joseph"},
#41	{"role": "assistant", "content": "Hello Joseph, it's nice to meet you!"},
#42	{"role": "user", "content": "I'm from Seattle and I work as a software engineer"}
#43	]
#44
#45	# Enable graph memory when adding
#46	client.add(
#47	messages,
#48	user_id="joseph",
#49	enable_graph=True
#50	)
#51	```
#52
#53	```javascript JavaScript
#54	import { MemoryClient } from "mem0";
#55
#56	const client = new MemoryClient({
#57	apiKey: "your-api-key",
#58	org_id: "your-org-id",
#59	project_id: "your-project-id"
#60	});
#61
#62	const messages = [
#63	{ role: "user", content: "My name is Joseph" },
#64	{ role: "assistant", content: "Hello Joseph, it's nice to meet you!" },
#65	{ role: "user", content: "I'm from Seattle and I work as a software engineer" }
#66	];
#67
#68	// Enable graph memory when adding
#69	await client.add({
#70	messages,
#71	user_id: "joseph",
#72	enable_graph: true
#73	});
#74	```
#75
#76	```json Output
#77	{
#78	"results": [
#79	{
#80	"memory": "Name is Joseph",
#81	"event": "ADD",
#82	"id": "4a5a417a-fa10-43b5-8c53-a77c45e80438"
#83	},
#84	{
#85	"memory": "Is from Seattle",
#86	"event": "ADD",
#87	"id": "8d268d0f-5452-4714-b27d-ae46f676a49d"
#88	},
#89	{
#90	"memory": "Is a software engineer",
#91	"event": "ADD",
#92	"id": "5f0a184e-ddea-4fe6-9b92-692d6a901df8"
#93	}
#94	]
#95	}
#96	```
#97	</CodeGroup>
#98
#99	The graph memory would look like this:
#100
#101	<Frame>
#102	<img src="/images/graph-platform.png" alt="Graph Memory Visualization showing relationships between entities" />
#103	</Frame>
#104
#105	<Caption>Graph Memory creates a network of relationships between entities, enabling more contextual retrieval</Caption>
#106
#107
#108	<Note>
#109	Response for the graph memory's `add` operation will not be available directly in the response. As adding graph memories is an asynchronous operation due to heavy processing, you can use the `get_all()` endpoint to retrieve the memory with the graph metadata.
#110	</Note>
#111
#112
#113	### Searching with Graph Memory
#114
#115	When searching memories, Graph Memory helps retrieve entities that are contextually important even if they're not direct semantic matches.
#116
#117	<CodeGroup>
#118
#119	```python Python
#120	# Search with graph memory enabled
#121	results = client.search(
#122	"what is my name?",
#123	user_id="joseph",
#124	enable_graph=True
#125	)
#126
#127	print(results)
#128	```
#129
#130	```javascript JavaScript
#131	// Search with graph memory enabled
#132	const results = await client.search({
#133	query: "what is my name?",
#134	user_id: "joseph",
#135	enable_graph: true
#136	});
#137
#138	console.log(results);
#139	```
#140
#141	```json Output
#142	{
#143	"results": [
#144	{
#145	"id": "4a5a417a-fa10-43b5-8c53-a77c45e80438",
#146	"memory": "Name is Joseph",
#147	"user_id": "joseph",
#148	"metadata": null,
#149	"categories": ["personal_details"],
#150	"immutable": false,
#151	"created_at": "2025-03-19T09:09:00.146390-07:00",
#152	"updated_at": "2025-03-19T09:09:00.146404-07:00",
#153	"score": 0.3621795393335552
#154	},
#155	{
#156	"id": "8d268d0f-5452-4714-b27d-ae46f676a49d",
#157	"memory": "Is from Seattle",
#158	"user_id": "joseph",
#159	"metadata": null,
#160	"categories": ["personal_details"],
#161	"immutable": false,
#162	"created_at": "2025-03-19T09:09:00.170680-07:00",
#163	"updated_at": "2025-03-19T09:09:00.170692-07:00",
#164	"score": 0.31212713194651254
#165	}
#166	],
#167	"relations": [
#168	{
#169	"source": "joseph",
#170	"source_type": "person",
#171	"relationship": "name",
#172	"target": "joseph",
#173	"target_type": "person",
#174	"score": 0.39
#175	}
#176	]
#177	}
#178	```
#179
#180	</CodeGroup>
#181
#182	<Note>
#183	`results` always reflects the vector search order (optionally reranked). Graph Memory augments that response by adding related entities in the `relations` array; it does not re-rank the vector results automatically.
#184	</Note>
#185
#186	### Retrieving All Memories with Graph Memory
#187
#188	When retrieving all memories, Graph Memory provides additional relationship context:
#189
#190	<Callout type="warning" title="Filters Required">
#191	`get_all()` now requires filters to be specified.
#192	</Callout>
#193
#194	<CodeGroup>
#195
#196	```python Python
#197	# Get all memories with graph context
#198	memories = client.get_all(
#199	filters={"AND": [{"user_id": "joseph"}]},
#200	enable_graph=True
#201	)
#202
#203	print(memories)
#204	```
#205
#206	```javascript JavaScript
#207	// Get all memories with graph context
#208	const memories = await client.getAll({
#209	filters: {"AND": [{"user_id": "joseph"}]},
#210	enable_graph: true
#211	});
#212
#213	console.log(memories);
#214	```
#215
#216	```json Output
#217	{
#218	"results": [
#219	{
#220	"id": "5f0a184e-ddea-4fe6-9b92-692d6a901df8",
#221	"memory": "Is a software engineer",
#222	"user_id": "joseph",
#223	"metadata": null,
#224	"categories": ["professional_details"],
#225	"immutable": false,
#226	"created_at": "2025-03-19T09:09:00.194116-07:00",
#227	"updated_at": "2025-03-19T09:09:00.194128-07:00",
#228	},
#229	{
#230	"id": "8d268d0f-5452-4714-b27d-ae46f676a49d",
#231	"memory": "Is from Seattle",
#232	"user_id": "joseph",
#233	"metadata": null,
#234	"categories": ["personal_details"],
#235	"immutable": false,
#236	"created_at": "2025-03-19T09:09:00.170680-07:00",
#237	"updated_at": "2025-03-19T09:09:00.170692-07:00",
#238	},
#239	{
#240	"id": "4a5a417a-fa10-43b5-8c53-a77c45e80438",
#241	"memory": "Name is Joseph",
#242	"user_id": "joseph",
#243	"metadata": null,
#244	"categories": ["personal_details"],
#245	"immutable": false,
#246	"created_at": "2025-03-19T09:09:00.146390-07:00",
#247	"updated_at": "2025-03-19T09:09:00.146404-07:00",
#248	}
#249	],
#250	"relations": [
#251	{
#252	"source": "joseph",
#253	"source_type": "person",
#254	"relationship": "name",
#255	"target": "joseph",
#256	"target_type": "person"
#257	},
#258	{
#259	"source": "joseph",
#260	"source_type": "person",
#261	"relationship": "city",
#262	"target": "seattle",
#263	"target_type": "city"
#264	},
#265	{
#266	"source": "joseph",
#267	"source_type": "person",
#268	"relationship": "job",
#269	"target": "software engineer",
#270	"target_type": "job"
#271	}
#272	]
#273	}
#274	```
#275
#276	</CodeGroup>
#277
#278	### Setting Graph Memory at Project Level
#279
#280	Instead of passing `enable_graph=True` to every add call, you can enable it once at the project level:
#281
#282	<CodeGroup>
#283
#284	```python Python
#285	from mem0 import MemoryClient
#286
#287	client = MemoryClient(
#288	api_key="your-api-key",
#289	org_id="your-org-id",
#290	project_id="your-project-id"
#291	)
#292
#293	# Enable graph memory for all operations in this project
#294	client.project.update(enable_graph=True)
#295
#296	# Now all add operations will use graph memory by default
#297	messages = [
#298	{"role": "user", "content": "My name is Joseph"},
#299	{"role": "assistant", "content": "Hello Joseph, it's nice to meet you!"},
#300	{"role": "user", "content": "I'm from Seattle and I work as a software engineer"}
#301	]
#302
#303	client.add(
#304	messages,
#305	user_id="joseph"
#306	)
#307	```
#308
#309	```javascript JavaScript
#310	import { MemoryClient } from "mem0";
#311
#312	const client = new MemoryClient({
#313	apiKey: "your-api-key",
#314	org_id: "your-org-id",
#315	project_id: "your-project-id"
#316	});
#317
#318	// Enable graph memory for all operations in this project
#319	await client.project.update({ enable_graph: true });
#320
#321	// Now all add operations will use graph memory by default
#322	const messages = [
#323	{ role: "user", content: "My name is Joseph" },
#324	{ role: "assistant", content: "Hello Joseph, it's nice to meet you!" },
#325	{ role: "user", content: "I'm from Seattle and I work as a software engineer" }
#326	];
#327
#328	await client.add({
#329	messages,
#330	user_id: "joseph"
#331	});
#332	```
#333
#334	</CodeGroup>
#335
#336
#337	## Best Practices
#338
#339	- Enable Graph Memory for applications where understanding context and relationships between memories is important.
#340	- Graph Memory works best with a rich history of related conversations.
#341	- Consider Graph Memory for long-running assistants that need to track evolving information.
#342
#343	## Performance Considerations
#344
#345	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.
#346
#347	If you have any questions, please feel free to reach out to us using one of the following methods:
#348
#349	<Snippet file="get-help.mdx" />
#350

z6Mkq5mY3JWtxoxUobWcfNHm7AkRubgSWEZTkBVqZXJviFZ5/my-project-public