Ioanna Lytra

Bio: Ioanna Lytra is an academic researcher from University of Bonn. The author has contributed to research in topics: Question answering & Architectural pattern. The author has an hindex of 11, co-authored 35 publications receiving 379 citations. Previous affiliations of Ioanna Lytra include Fraunhofer Society & University of Vienna.

Why Reinvent the Wheel: Let's Build Question Answering Systems Together

Abstract: 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 the effectiveness of the pipelines generated by Frankenstein using the QALD and LC-QuAD benchmarks. These results not only suggest that Frankenstein precisely solves the QA optimisation problem but also enables the automatic composition of optimised QA pipelines, which outperform the static Baseline QA pipeline. Thanks to this flexible and fully automated pipeline generation process, new QA components can be easily included in Frankenstein, thus improving the performance of the generated pipelines.

MULDER: Querying the Linked Data Web by Bridging RDF Molecule Templates

Abstract: The increasing number of RDF data sources that allow for querying Linked Data via Web services form the basis for federated SPARQL query processing. Federated SPARQL query engines provide a unified view of a federation of RDF data sources, and rely on source descriptions for selecting the data sources over which unified queries will be executed. Albeit efficient, existing federated SPARQL query engines usually ignore the meaning of data accessible from a data source, and describe sources only in terms of the vocabularies utilized in the data source. Lack of source description may conduce to the erroneous selection of data sources for a query, thus affecting the performance of query processing over the federation. We tackle the problem of federated SPARQL query processing and devise MULDER, a query engine for federations of RDF data sources. MULDER describes data sources in terms of RDF molecule templates, i.e., abstract descriptions of entities belonging to the same RDF class. Moreover, MULDER utilizes RDF molecule templates for source selection, and query decomposition and optimization. We empirically study the performance of MULDER on existing benchmarks, and compare MULDER performance with state-of-the-art federated SPARQL query engines. Experimental results suggest that RDF molecule templates empower MULDER federated query processing, and allow for the selection of RDF data sources that not only reduce execution time, but also increase answer completeness.

Capturing Knowledge in Semantically-typed Relational Patterns to Enhance Relation Linking

Abstract: Transforming natural language questions into formal queries is an integral task in Question Answering (QA) systems. QA systems built on knowledge graphs like DBpedia, require a step after natural language processing for linking words, specifically including named entities and relations, to their corresponding entities in a knowledge graph. To achieve this task, several approaches rely on background knowledge bases containing semantically-typed relations, e.g., PATTY, for an extra disambiguation step. Two major factors may affect the performance of relation linking approaches whenever background knowledge bases are accessed: a) limited availability of such semantic knowledge sources, and b) lack of a systematic approach on how to maximize the benefits of the collected knowledge. We tackle this problem and devise SIBKB, a semantic-based index able to capture knowledge encoded on background knowledge bases like PATTY. SIBKB represents a background knowledge base as a bi-partite and a dynamic index over the relation patterns included in the knowledge base. Moreover, we develop a relation linking component able to exploit SIBKB features. The benefits of SIBKB are empirically studied on existing QA benchmarks and observed results suggest that SIBKB is able to enhance the accuracy of relation linking by up to three times.

Supporting consistency between architectural design decisions and component models through reusable architectural knowledge transformations

Abstract: In recent years, the software architecture community has proposed to use architectural design decisions (ADDs) for capturing the design rationale and the architectural knowledge (AK) As software systems evolve both ADDs and architectural designs need to be documented and maintained This is a tedious and time-consuming task because of the lack of systematic and automated support for bridging between ADDs and designs As a result, decisions and designs become inconsistent over time We propose to alleviate this problem by introducing an AK transformation language supporting reusable AK transformations from pattern-based ADDs to component-and-connector models In addition, we devise reusable consistency checking rules for verifying the consistency between decisions and designs Through the use of model-driven transformations, as well as reusable, pattern-based decision models, we ensure the reusability of our approach We apply our approach in an industrial case study and show that it offers high reusability, is largely automated and scalable, and can deal with the complexity of large numbers of recurring decisions

Architectural Decision Making for Service-Based Platform Integration: A Qualitative Multi-Method Study

Abstract: Nowadays the software architecture of a system is often seen as a set of design decisions providing the rationale for the system design. When designing a software architecture multiple levels of design decisions need to be considered. For example, the service-based integration of heterogeneous platforms and the development of applications on top of those integration services requires high-level as well as technology-, domain-, and application-dependent architectural decisions. In this context, we performed a series of qualitative studies following a multi-method approach. First, we conducted a systematic literature review from which we derived a pattern language for platform integration featuring 40 patterns, as well as a pattern-based architectural decision model. Then, we performed interviews with 9 platform experts from 3 companies for revising the architectural knowledge captured by the pattern language and the decision model. Finally, we participated in a case study and observed the decision-making process to validate the results further. Our observations resulted in 1) a qualitatively validated, pattern-based architectural decision model and 2) a generalized model of the different levels and stages of architectural decision making for service-based platform integration.

To the Best of Our Knowledge

Abstract: To the Editor.— The editorial by Samuel Vaisrub, MD, "To the Best of Our Knowledge, Which Is Limited at Best" (241:278, 1979), was delightful, and its point was unhappily correct. It brought to my mind that demolisher of analytical studies, "to the best of our knowledge," or "as far as we know." It goes like this: We treated so many people, and we know this many are well, and so many have died; and this many,from whom we have not heard, are alive and well,as far as we know. Someone from whom you have not heard is more likely to be dead than alive.

Software product line engineering and variability management: achievements and challenges

Abstract: Software product line engineering has proven to empower organizations to develop a diversity of similar software-intensive systems (applications) at lower cost, in shorter time, and with higher quality when compared with the development of single systems. Over the last decade the software product line engineering research community has grown significantly. It has produced impressive research results both in terms of quality as well as quantity. We identified over 600 relevant research and experience papers published within the last seven years in established conferences and journals. We briefly summarize the major research achievements of these past seven years. We structure this research summary along a standardized software product line framework. Further, we outline current and future research challenges anticipated from major trends in software engineering and technology.