The QALL-ME Framework: A specifiable-domain multilingual Question Answering architecture

Ferrández, Óscar; Spurk, Christian; Kouylekov, Milen; Dornescu, Iustin; Ferrández, Sergio; Negri, Matteo; Izquierdo, Rubén; Tomás, David; Orăsan, Constantin; Neumann, Guenter; Magnini, Bernardo; Vicedo, Jose-Luis

doi:10.1016/j.websem.2011.01.002

Cited by 45 publications

(19 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Since developing KGQA systems requires solving several tasks, namely entity, relation and class linking, and afterwards query building, they are often implemented as independent components and arranged into a single pipeline [9]. Frameworks such as QALL-ME [11], OKBQA [24] and Frankenstein [41], allow one to share and reuse those components as a collaborative effort. For example, Frankenstein includes 29 components that can be combined and interchanged [43].…”

Section: Related Workmentioning

confidence: 99%

Message Passing for Complex Question Answering over Knowledge Graphs

Vakulenko

García

Polleres

et al. 2019

Proceedings of the 28th ACM International Conference on Information and Knowledge Management

View full text Add to dashboard Cite

Question answering over knowledge graphs (KGQA) has evolved from simple single-fact questions to complex questions that require graph traversal and aggregation. We propose a novel approach for complex KGQA that uses unsupervised message passing, which propagates confidence scores obtained by parsing an input question and matching terms in the knowledge graph to a set of possible answers. First, we identify entity, relationship, and class names mentioned in a natural language question, and map these to their counterparts in the graph. Then, the confidence scores of these mappings propagate through the graph structure to locate the answer entities. Finally, these are aggregated depending on the identified question type.This approach can be efficiently implemented as a series of sparse matrix multiplications mimicking joins over small local subgraphs. Our evaluation results show that the proposed approach outperforms the state-of-the-art on the LC-QuAD benchmark. Moreover, we show that the performance of the approach depends only on the quality of the question interpretation results, i.e., given a correct relevance score distribution, our approach always produces a correct answer ranking. Our error analysis reveals correct answers missing from the benchmark dataset and inconsistencies in the DBpedia knowledge graph. Finally, we provide a comprehensive evaluation of the proposed approach accompanied with an ablation study and an error analysis, which showcase the pitfalls for each of the question answering components in more detail.

show abstract

Section: Related Workmentioning

confidence: 99%

Message Passing for Complex Question Answering over Knowledge Graphs

Vakulenko

García

Polleres

et al. 2019

Proceedings of the 28th ACM International Conference on Information and Knowledge Management

View full text Add to dashboard Cite

show abstract

“…Собранная с помощью Qanary-фреймворка система может быть интегрирована в другие системы как Springсервис. В работе [11] представлен фреймворк QALL-ME, предоставляющий переиспользуемую архитектуру для создания мультиязычных вопросно-ответных систем, учитывающих пространственно-временной контекст запроса. Авторы декомпозируют процесс трансформации на этапы и сопоставляют с каждым этапом отдельный тип компонентов, например аннотаторы именованных сущностей для различных языков.…”

Section: обзор существующих платформ создания средств трансформацииunclassified

Platform for Integrating and Testing Tools which Transform Natural Language Queries into SPARQL Queries

Postanogov

Turova

2019

JIT

View full text Add to dashboard Cite

In the paper we discuss how to support the process of creating tools which transform natural language (NL) queries into SPARQL queries (hereinafter referred to as a transformation tool). In the introduction, we describe the relevance of the task of understanding natural language queries by information systems, as well as the advantages of using ontologies as a means of representing knowledge for solving this problem. This ontology-based data access approach can be also used in systems which provide natural language interface to databases. Based on the analysis of problems related to the integration and testing of existing transformation tools, as well as to support the creation and testing own transformation modules, the concept of a software platform that simplifies these tasks is proposed. The platform architecture satisfies the requirements for ease of connecting third party transformation tools, reusing individual modules, as well as integrating the resulting transformation tools into other systems, including testing systems. The building blocks of the created transformation systems are the individual transformation modules packaged in Docker containers. Program access to each module is carried out using gRPC. Modules loaded into the platform can be built into the transformation pipeline automatically or manually using the built-in third party SciVi data flow diagram editor. Compatibility of individual modules is controlled by automatic analysis of application programming interfaces. The resulting pipeline is combined according to specified data flow into a single multi-container application that can be integrated into other systems, as well as tested on extendable test suites. The expected and actual results of the query transformation are available for viewing in graphical form in the visualization tool developed earlier.

show abstract

“…If so, the generated QA pipeline is added to the algorithm output. In Equation 5, the algorithm finds that only the QA pipeline Dandelion, ReMatch, and SINA produces results; the other two pipelines fail because the QA components RNLIWOD 8 and Spot Property 9 are not able to perform the relation linking task of the question q="What is the capital of Canada?". The algorithm ends when the top k i QA components have been combined and checked; the output is the union of the best performing QA pipelines that produce a non-empty answer.…”

Section: Identifying Optimal Qa Pipelinesmentioning

confidence: 99%

“…7 https://dandelion.eu/docs/api/datatxt/nex/getting-started/ 8 Component based on https://github.com/dice-group/NLIWOD. 9 This component is the combination of the NLIWOD and RL components of [15].…”

Section: Identifying Optimal Qa Pipelinesmentioning

confidence: 99%

Why Reinvent the Wheel

Singh

Radhakrishna

Both

et al. 2018

Proceedings of the 2018 World Wide Web Conference on World Wide Web - WWW '18

View full text Add to dashboard Cite

Modern question answering (QA) systems need to flexibly integrate a number of components specialised to fulfil specific tasks in a QA pipeline. Key QA tasks include Named Entity Recognition and Disambiguation, Relation Extraction, and Query Building. Since a number of different software components exist that implement different strategies for each of these tasks, it is a major challenge to select and combine the most suitable components into a QA system, given the characteristics of a question. We study this optimisation problem and train classifiers, which take features of a question as input and have the goal of optimising the selection of QA components based on those features. We then devise a greedy algorithm to identify the pipelines that include the suitable components and can effectively answer the given question. We implement this model within Frankenstein, a QA framework able to select QA components and compose QA pipelines. We evaluate This paper is published under the Creative Commons Attribution 4.0 International (CC BY 4.0) license. Authors reserve their rights to disseminate the work on their personal and corporate Web sites with the appropriate attribution.

show abstract

The QALL-ME Framework: A specifiable-domain multilingual Question Answering architecture

Cited by 45 publications

References 13 publications

Message Passing for Complex Question Answering over Knowledge Graphs

Message Passing for Complex Question Answering over Knowledge Graphs

Platform for Integrating and Testing Tools which Transform Natural Language Queries into SPARQL Queries

Why Reinvent the Wheel

Contact Info

Product

Resources

About