Spatio‐temporal pseudo relevance feedback for scientific data retrieval

Abstract

We consider the problem of searching scientific data from vast heterogeneous scientific data repositories. This problem is challenging because scientific data contain relatively little text information compared to other search targets such as web pages. On the other hand, the metadata in scientific data contain other characteristic information such as spatio‐temporal information. Although using this information make it possible to improve the search performance, many widely adopted scientific data search engines use this information exclusively for narrowing down search results. In this paper, we propose a novel query generation method using spatial, temporal, and text information based on pseudo relevance feedback. The proposed method generates new spatio‐temporal queries from the initial search results. By using these queries, the search results are reranked such that more related results obtain higher rank. The experimental results show that the proposed method outperforms a baseline method when search targets do not have rich text information. © 2016 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

Word-embedding-based query expansion: Incorporating Deep Averaging Networks in Arabic document retrieval

“…WordNet is typically the most common external tool used by global methods to select new candidate expansion terms that are semantically associated with the original query terms [7]. In contrast, local methods use relevance feedback, in which results from an initial retrieval are used to select the most promising terms to be added to the original query [8,9]. Pseudo Relevance Feedback (PRF), which assumes that the top-ranking k documents from the initial retrieval are relevant for selecting new terms, is a more favourable expansion technique as it automates manual relevance feedback.…”

Patterns for searching data on the web across different research communities

“…Several works have addressed similarities and differences between document and data search (Kern & Mathiak, 2015;Stempfhuber & Zapilko, 2009), emphasizing the need for new approaches and solutions with respect to the latter. As information objects, data are different from documents, for they are primarily numerical and non-textual (Takeuchi, Sugiura, Akahoshi, & Zettsu, 2017). Kern and Mathiak (2015) observed that users put much more effort in detailed querying of data, expressing locations, ranges and operators at an average word count of nine.…”

“…Another approach would be fostering the discovery of datasets according to their spatial and temporal properties. Despite their heterogeneity, inconsistency and incompleteness across different research communities and their corresponding repositories, around 73% of metadata records, e.g., from the PANGAEA repository hold information on spatial and/or temporal information as two basic characteristics of data (Takeuchi et al, 2017). According to DataCite's metadata schema 4.3, geolocation has become a mandatory metadata element with the properties for specifying both the longitude/latitude and the label of a geolocation.…”