How-to

Oracle AI Agent Memory 26.8 引入图感知检索

Why we picked it

作者写了个具体例子演示图检索:把事实存成记忆记录,再用 supersedes、contradicts 这类关系标签连起来检索。做智能体记忆或知识库的朋友可以对照 Oracle AI Agent Memory 26.8 的博客上手试,比纯向量检索更适合维护会被更新、推翻的事实。

Oracle 在 AI Agent Memory 26.8 版本中实现了 Graph-aware search(图感知检索)。具体做法分三步:先把事实存为持久化记忆记录,再用标注关系类型的链接把记录连起来,例如 supersedes(取代)、supports(支持)、contradicts(矛盾),最后检索时利用这些关系链接找到正确的记录。这解决了纯语义检索的短板:向量相似度只看文本含义,无法表达事实之间的逻辑关系。

Body

对需要在事实之间维护关系的应用场景,比如智能体记忆、知识库中信息被更新或推翻的情况,图感知检索可以按关系直接定位,而不是靠语义匹配碰运气。作者提到这是与 Oracle Devs 团队合作撰写的博客,附有完整实现示例,Oracle 的开发者可以直接在新版本中试用这一能力。

Original · Ate-a-Pi

Graph-aware search beats semantic search every day of the week and twice on Sunday.

I spent all night putting together an example to show you how cool graph retrieval is.

1. You store facts as persistent memory records.

2. You connect records with links labeled with their relationships. For example, "supersedes," "supports," or "contradicts".

3. You then use these links to retrieve the proper records.

If you need to store information with relationships between facts, you want to look into this.

Oracle AI Agent Memory 26.8 implements graph-aware search.

Here is a blog post explaining how Graph-aware search works in Oracle:

https://t.co/OhGOuR6aZG

Thanks to the @OracleDevs team for partnering with me on this post.

Industry Coordinates & Source Verification

This report is categorized under AITOP's "How-to" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Oracle, RAG, agents, MCP and tools.

Source attribution: curated from Ate-a-Pi (@svpino).