Advanced Semantics for Commonsense Knowledge Extraction (2020-11-02T00:00:00.000000Z)

TL;DR

This paper presents a methodology, called Ascent, to automatically build a large-scale knowledge base (KB) of CSK assertions, with advanced expressiveness and both better precision and recall than prior works.

Abstract

Commonsense knowledge (CSK) about concepts and their properties is useful for AI applications such as robust chatbots. Prior works like ConceptNet, TupleKB and others compiled large CSK collections, but are restricted in their expressiveness to subject-predicate-object (SPO) triples with simple concepts for S and monolithic strings for P and O. Also, these projects have either prioritized precision or recall, but hardly reconcile these complementary goals. This paper presents a methodology, called Ascent, to automatically build a large-scale knowledge base (KB) of CSK assertions, with advanced expressiveness and both better precision and recall than prior works. Ascent goes beyond triples by capturing composite concepts with subgroups and aspects, and by refining assertions with semantic facets. The latter are important to express temporal and spatial validity of assertions and further qualifiers. Ascent combines open information extraction with judicious cleaning using language models. Intrinsic evaluation shows the superior size and quality of the Ascent KB, and an extrinsic evaluation for QA-support tasks underlines the benefits of Ascent. A web interface, data and code can be found at https://www.mpi-inf.mpg.de/ascent.

Authors

G. Weikum

14 papers

Simon Razniewski

3 papers

Tuan-Phong Nguyen

2 papers

TL;DR

Abstract

Authors

References69 items

Commonsense Knowledge in Wikidata

Do Dogs have Whiskers? A New Knowledge Base of hasPart Relations

Language Models are Few-Shot Learners

GenericsKB: A Knowledge Base of Generic Statements

TransOMCS: From Linguistic Graphs to Commonsense Knowledge

Probing Neural Language Models for Human Tacit Assumptions

GIANT: Scalable Creation of a Web-scale Ontology

Knowledge Graphs

How Context Affects Language Models' Factual Predictions

REALM: Retrieval-Augmented Language Model Pre-Training

Joint Reasoning for Multi-Faceted Commonsense Knowledge

ALBERT: A Lite BERT for Self-supervised Learning of Language Representations

Uncertain Natural Language Inference

KagNet: Knowledge-Aware Graph Networks for Commonsense Reasoning

Language Models as Knowledge Bases?

Incorporating Relation Knowledge into Commonsense Reading Comprehension with Multi-task Learning

RoBERTa: A Robustly Optimized BERT Pretraining Approach

How Large Are Lions? Inducing Distributions over Quantitative Attributes

Commonsense Properties from Query Logs and Question Answering Forums

StuffIE: Semantic Tagging of Unlabeled Facets Using Fine-Grained Information Extraction

OntoSenticNet: A Commonsense Ontology for Sentiment Analysis

Graphene: Semantically-Linked Propositions in Open Information Extraction

A Survey on Open Information Extraction

Supervised Open Information Extraction

Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge

Olive Oil is Made of Olives, Baby Oil is Made for Babies: Interpreting Noun Compounds Using Paraphrases in a Neural Model

WebIsALOD: Providing Hypernymy Relations Extracted from the Web as Linked Open Data

Reasoning with Heterogeneous Knowledge for Commonsense Machine Comprehension

Domain-Targeted, High Precision Knowledge Extraction

Ordinal Common-sense Inference

Open Information Extraction Systems and Downstream Applications

A Large DataBase of Hypernymy Relations Extracted from the Web.

Commonsense in Parts: Mining Part-Whole Relations from the Web and Image Tags

WebChild: harvesting and organizing commonsense knowledge from the web

The Centre for Speech, Language and the Brain (CSLB) concept property norms

Reporting bias and knowledge acquisition

Distributed Representations of Words and Phrases and their Compositionality

BabelNet: The automatic construction, evaluation and application of a wide-coverage multilingual semantic network

Probase: a probabilistic taxonomy for text understanding

Mapping WordNet synsets to Wikipedia articles

Taxonomy induction based on a collaboratively built knowledge repository

Quantificational Sharpening of Commonsense Knowledge

Semantic Role Labeling

Word sense disambiguation: A survey

Weakly-Supervised Acquisition of Open-Domain Classes and Class Attributes from Web Documents and Query Logs

Semantic Taxonomy Induction from Heterogenous Evidence

Espresso: Leveraging Generic Patterns for Automatically Harvesting Semantic Relations

Automatic Discovery of Part-Whole Relations

Web-scale information extraction in knowitall: (preliminary results)

Many-Dimensional Modal Logics: Theory and Applications

Open Mind Common Sense: Knowledge Acquisition from the General Public

Can we derive general world knowledge from texts

CYC: a large-scale investment in knowledge infrastructure

WordNet: A Lexical Database for English

Programs with common sense

BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

Knowledge Engineering

Synthesis Lectures on Human Language Technologies

ConceptNet 5: A Large Semantic Network for Relational Knowledge

An NLP Curator (or: How I Learned to Stop Worrying and Love NLP Pipelines)

Learning from the Web: Extracting General World Knowledge from Noisy Text

Reasoning

The original algorithm frequently returns assertions with empty objects. To only return complete triples

Carsusediesel.Carsusefuel.Carsusecircuitbreakers.Carsusetransmission.Cars useengines.Carsusepropa... Carsusedforthepurposeoftrans-portinggoods

Assertions that have a subject among the lemmas of the main subject, and an object which is a noun chunk consisting of an aspect t ∈ asp s 0 as the head noun and an adjectival modifier

We post-process special cases of sentences used for giving examples with the words: “like” , “such

Carsarecomplexdevicesusedin acomplexen-vironment

each assertion 𝑎 𝑖

LMs are provided with an answer sentence prefix

Field of Study

Journal Information

Name

Venue Information

Name

Type

URL