首页 > AI前沿 > AutoOntic: Automatic Ontology Induction from Text for Knowledge Graph Construction

AutoOntic: Automatic Ontology Induction from Text for Knowledge Graph Construction

arXiv自然语言 2026-11-11 21:37 30 阅读 查看原文
用户现在需要我把给定的英文新闻正文排版成语义化HTML片段,首先看原文都是自然段落,没有小节所以不用h2/h3,每个段落用p包裹?不对,原文是一整段?不对,仔细看原文:开头是讲Graph-based RAG的问题,然后介绍AutoOntic,然后它的方法,然后评估,然后结果,最后可用链接。不对,原文是一个连贯的段落?不对,等下看原文内容:首先第一部分是背景,然后“We present AutoOntic...”是新的段落?不对,用户给的正文是一整段吗?不对,仔细看,用户给的正文是: 第一句:Graph-based retrieval-augmented generation depends on the quality of the knowledge graph (KG) it retrieves over, yet schema-free LLM-based extraction pipelines typically produce flat graphs: deduplicated entities and relations with no type hierarchy and no domain/range constraints. 然后第二句开始:We present AutoOntic, a pipeline that automatically constructs an ontology from text by canonicalizing open-vocabulary entity types and relations, inducing a hierarchy over the canonicalized types, and mining relation-level domain and range constraints. 这部分是介绍AutoOntic是什么,是不是要分成段落?不对,规则说每个自然段落用p包裹,原文的自然段落是怎么划分的?哦,用户给的正文看起来是一个自然段落?不对,不对,再仔细看,原文是连续的文本,但可以按语义划分段落吗?不对,规则说保留原文所有信息,不得增删改内容,那首先,原文的内容要完整保留,然后如果有需要强调的可以加strong?比如AutoOntic可以加strong?还有重要的结果比如88%、over 90%可以加strong?然后最后那个链接是地址,不用特殊处理? 等下,原文的结构: 1. 背景句:Graph-based...constraints. 2. 介绍AutoOntic的句:We present AutoOntic...constraints. 3. 它的方法相关:Its hierarchy induction procedure...links. 4. 作用:The resulting hierarchy...constraints. 5. 评估方法:For intrinsic evaluation...classes. 6. 实验结果1:On the Wikidata-TekGen...granularity. 7. 实验结果2:On MINE...accuracy. 8. 可用地址:AutoOntic is available at...AutoOntic. 这些是自然的逻辑段落吗?还是原文是一个整段?不对,用户给的正文是一段连续的文本,但按照排版规则,每个自然段落用p包裹,那我应该把语义相关的分成段落?不对,等下,用户给的原文是不是本身就是多个自然段落?不,看用户给的正文是连在一起的,没有换行。哦,不对,可能我误解了,原文的纯文本是一段,但可以按语义划分成多个p吗?不对,规则说保留原文所有信息,不得增删改,所以内容不能改,但是段落划分是不是可以按语义?比如: 首先

里放背景部分:Graph-based retrieval-augmented generation depends on the quality of the knowledge graph (KG) it retrieves over, yet schema-free LLM-based extraction pipelines typically produce flat graphs: deduplicated entities and relations with no type hierarchy and no domain/range constraints.

然后

里放介绍AutoOntic的:We present AutoOntic, a pipeline that automatically constructs an ontology from text by canonicalizing open-vocabulary entity types and relations, inducing a hierarchy over the canonicalized types, and mining relation-level domain and range constraints.

然后下一个

讲它的层次归纳过程:Its hierarchy induction procedure prioritizes types using corpus-derived relation-argument profiles and places them through local, LLM-adjudicated parent/child checks, allowing newly inserted types to trigger reassignment of existing parent-child links.

然后下一个

讲结果的作用:The resulting hierarchy constrains entity deduplication and supports the induction of relation-level domain/range constraints.

然后下一个

讲评估方法:For intrinsic evaluation, we introduce a reference-based protocol that classifies induced ontology types as exact, broader, narrower, or unrelated to reference domain and range classes.

然后下一个

讲第一个实验结果:On the Wikidata-TekGen and DBpedia-WebNLG subsets of Text2KGBench, over 90% of AutoOntic's induced endpoint types are semantically aligned with the reference classes across two LLM backbones: about half match exactly, while most remaining differences concern granularity.

然后下一个

讲第二个实验结果:On MINE, knowledge graphs organized using an ontology built by AutoOntic achieve 88% retrieval accuracy.

然后最后一个

讲可用地址:AutoOntic is available at https://github.com/screemix/AutoOntic.

不对,但是这样会不会改变了原文的段落?哦,原文的纯文本没有分段,那是不是应该保持成一个p?不对,等下看规则2:每个自然段落用

包裹,长文用h2/h3划分小节。原文是学术类的摘要,没有小节,所以不用h2/h3。那如果原文是一个自然段落,就用一个p?但这样排版不太语义化?不对,再仔细看原文的逻辑,其实是连贯的,但可以按语义拆分段落吗?不对,规则说不得增删改内容,但是段落划分是排版的一部分,不算改内容。哦,对,排版的时候可以把逻辑上不同的部分分成不同的p,只要内容不变。 等下,再检查有没有需要用其他标签的,比如没有列表、引用、代码,所以不用ul、blockquote、pre。然后重要的句子或者关键词用strong,比如AutoOntic是重要的,还有关键数据over 90%、88%可以加strong。 现在整理一下,确保所有内容都在,没有增删: 首先,第一个p放背景:

Graph-based retrieval-augmented generation depends on the quality of the knowledge graph (KG) it retrieves over, yet schema-free LLM-based extraction pipelines typically produce flat graphs: deduplicated entities and relations with no type hierarchy and no domain/range constraints.

然后第二个p介绍AutoOntic:

We present AutoOntic, a pipeline that automatically constructs an ontology from text by canonicalizing open-vocabulary entity types and relations, inducing a hierarchy over the canonicalized types, and mining relation-level domain and range constraints.

第三个p讲层次归纳过程:

Its hierarchy induction procedure prioritizes types using corpus-derived relation-argument profiles and places them through local, LLM-adjudicated parent/child checks, allowing newly inserted types to trigger reassignment of existing parent-child links.

第四个p讲层次的作用:

The resulting hierarchy constrains entity deduplication and supports the induction of relation-level domain/range constraints.

第五个p讲评估方法:

For intrinsic evaluation, we introduce a reference-based protocol that classifies induced ontology types as exact, broader, narrower, or unrelated to reference domain and range classes.

第六个p讲第一个实验结果:

On the Wikidata-TekGen and DBpedia-WebNLG subsets of Text2KGBench, over 90% of AutoOntic's induced endpoint types are semantically aligned with the reference classes across two LLM backbones: about half match exactly, while most remaining differences concern granularity.

第七个p讲第二个实验结果:

On MINE, knowledge graphs organized using an ontology built by AutoOntic achieve 88% retrieval accuracy.

第八个p讲可用性:

AutoOntic is available at https://github.com/screemix/AutoOntic.

等下,但是有没有可能原文是一个整段,那是不是应该用一个p?不对,看规则说“每个自然段落用

包裹”,这里的自然段落是指原文中的自然段落,但原文的纯文本没有换行,是一段。那是不是应该用一个p?但这样排版不太好?不对,再仔细看用户给的正文,确实是连续的一段,没有换行,那是不是所有内容放在一个p里?但可以在里面加strong。比如:

Graph-based retrieval-augmented generation depends on the quality of the knowledge graph (KG