Knowledge Graphs Querying - Leitura de Artigo

Arijit Khan. 2023. Knowledge Graphs Querying. SIGMOD Rec. 52, 2 (June 2023), 18–29. https://doi.org/10.1145/3615952.3615956

ABSTRACT
Querying KGs is critical in web search, question answering (QA), semantic search, personal assistants, fact checking, and recommendation.

[Sistemas / tarefas onde consulta aos KGs é usada]

First, research on KG querying has been conducted by several communities, such as databases, data mining, semantic web, machine learning, information retrieval, and natural language processing (NLP), with different focus and terminologies; and also in diverse topics ranging from graph databases, query languages, join algorithms, graph patterns matching, to more sophisticated KG embedding and natural language questions (NLQs).

[Diversas perspectivas sobre os problemas que as consultas em KG trazem]

Second, many recent advances on KG and query embedding, multimodal KG, and KG-QA come from deep learning, IR, NLP, and computer vision domains. 

[De quais comunidades estão sendo propostas soluções e quais os problemas que ainda estão em aberto como por exemplo lidar com Incompletude]

1 Introduction

[KG para integração de fontes de dados com esquema flexível]

1.1 Challenges in KG Querying

Scalability and efficiency of graph query processing ....

Additionally, the notion of ‘relevant’ or ‘correct’ answers could very well depend on the user’s query intent, or can even be vague, thus a predefined, ‘one-size-fits-all’ similarity metric might not work in all scenarios.

[As abordagens de similaridade não dependem da tarefa]

Incomplete KGs. Knowledge graphs are incomplete and follow the open-world assumption — information
available in a KG only captures a subset of reality. To retrieve the complete set of correct answers for a given query, one must infer missing edges and relations.

[Completar o KG com dados do próprio KG ou de fontes externas] 

User-friendly querying

[Formulação interativa de consultas, completar, explicar]

2 Taxonomy of KG Querying

Graph workloads are broadly classified into two categories [67]: (1) online graph queries consisting of adhoc graph traversal and pattern matching – exploring a small fraction of the entire graph and requiring fast response time; (2) offline graph analytics with iterative, batch processing over the entire graph, e.g., PageRank, clustering, community detection, and machine learning algorithms.

The focus of this article is read-only online queries without updates in the KG. KG querying is essential for web search [129], QA [119], semantic search [155], personal assistants [12], fact checking [143], and recommendation [167].

[CaKQ Query Engine é online graph query] 

2.1 KG Data Models

[RDF e LPG]

2.2 KG Query and Question Classification

[Traduzir Pergunta em Consulta]

[Consulta simples ou complexa. Conjunção, Disjunção, Negação, etc ... Caminhos]

[Factoides x Agregada/Abstrata]

2.3 KG Query Languages & Technologies

[SPARQL, Cypher, GQL, Extensões para SQL, ...]

[Keyword]

2.4 Benchmarks for KG Query & QA

3 KG Query Processing & QA: Recent Neural Methods 

3.1 Embedding-based KG Query Processing

[Converter o KG e as consultas e achar a distância]

3.2 Multi-modal KG Embedding

3.3 Neural Methods for KG-QA

Answering natural language questions (NLQ) over knowledge graphs involve several subtasks including entity linking, relationships identification, identifying logical and numerical operators, query forms, intent, and finally the formal query construction [111]. Rule-based methods use ontologies and KG for phrase mapping and disambiguation to link entities and relations to the KG, and then employ grammars to generate formal queries.

Recently, neural network-based semantic parsing algorithms have become popular for KG-QA, which are categorized as classification, ranking, and translation-based [28]. 

3.4 Conversational QA on KG

4 Graph Databases Support for KG Query

5 Future Directions

Therefore, KGs can be a unified data model for complex data lake problems, to model cross-domain and diverse data.