Aggeliki Dimitriou

Top-k-size keyword search on tree structured data

Keyword search is the most popular technique for querying large tree-structured datasets, often of unknown structure, in the web. Recent keyword search approaches return lowest common ancestors (LCAs) of the keyword matches ranked with respect to their relevance to the keyword query. A major challenge of a ranking approach is the efficiency of its algorithms as the number of keywords and the size and complexity of the data increase. To face this challenge most of the known approaches restrict their ranking to a subset of the LCAs (e.g., SLCAs, ELCAs), missing relevant results.In this work, we design novel top-k-size stack-based algorithms on tree-structured data. Our algorithms implement ranking semantics for keyword queries which is based on the concept of LCA size. Similar to metric selection in information retrieval, LCA size reflects the proximity of keyword matches in the data tree. This semantics does not rank a predefined subset of LCAs and through a layered presentation of results, it demonstrates improved effectiveness compared to previous relevant approaches. To address performance challenges our algorithms exploit a lattice of the partitions of the keyword set, which empowers a linear time performance. This result is obtained without the support of auxiliary precomputed data structures. An extensive experimental study on various and large datasets confirms the theoretical analysis. The results show that, in contrast to other approaches, our algorithms scale smoothly when the size of the dataset and the number of keywords increase. HighlightsTop-k-size keyword search over tree-structured data.Efficient multi-stack threshold and top-k algorithms exploiting a lattice of keyword partitions.The algorithms demonstrate linear time complexity and scale smoothly when the size of the dataset increases.The approach does not rely on auxiliary indices and can be used in streaming applications.Novel layered ranking semantics based on the concept of LCA size.

XReason: A Semantic Approach That Reasons with Patterns to Answer XML Keyword Queries

Keyword search is a popular technique which allows querying multiple data sources on the web without having full knowledge of their structure. This flexibility comes with a drawback: usually, even though a large number of results match the user’s request only few of them are relevant to her intent. Since data on the web are often in tree-structured form, several approaches have been suggested in the past which attempt to exploit the structural properties of the data in order to filter out irrelevant results and return meaningful answers. This is certainly a difficult task, and depending on the type of dataset, these approaches show low precision and/or recall.

Exploiting Semantic Result Clustering to Support Keyword Search on Linked Data

Keyword search is by far the most popular technique for searching linked data on the web. The simplicity of keyword search on data graphs comes with at least two drawbacks: difficulty in identifying results relevant to the user intent among an overwhelming number of candidates and performance scalability problems. In this paper, we claim that result ranking and top-k processing which adapt schema unaware IR-based techniques to loosely structured data are not sufficient to address these drawbacks and efficiently produce answers of high quality. We present an alternative solution which hierarchically clusters the results based on a semantic interpretation of the keyword instances and takes advantage of relevance feedback from the user. Our clustering hierarchy exploits graph patterns which are structured queries clustering together result graphs of the same structure and represent possible interpretations for the keyword query. We present an algorithm which computes r-radius Steiner patterns graphs using exclusively the structural summary of the data graph. The user selects relevant pattern graphs by exploring only a small portion of the hierarchy supported by a ranking of the hierarchy components.Our experimental results show the feasibility of our system by demonstrating short reach times and efficient computation of the relevant results.

Reasoning with patterns to effectively answer XML keyword queries

Keyword search is a popular technique for searching tree-structured data on the Web because it frees the user from knowing a complex query language and the structure of the data sources. However, the imprecision of the keyword queries usually results in a very large number of results of which only a few are relevant to the query. Multiple previous approaches have tried to address this problem. They exploit the structural properties of the tree data in order to filter out irrelevant results. This is not an easy task though, and in the general case, these approaches show low precision and/or recall and low quality of result ranking. In this paper, we argue that exploiting the structural relationships of the query matches locally in the data tree is not sufficient and a global analysis of the keyword matches in the data tree is necessary in order to assign meaningful semantics to keyword queries. We present an original approach for answering keyword queries which extracts structural patterns of the query matches and reasons with them in order to return meaningful results ranked with respect to their relevance to the query. Comparisons between patterns are realized based on different types of homomorphisms between patterns. As the number of patterns is typically much smaller than that of the of query matches, this global reasoning is feasible. We design an efficient stack-based algorithm for evaluating keyword queries on tree-structured data, and we also devise a heuristic extension which further improves its performance. We run comprehensive experiments on different datasets to evaluate the efficiency of the algorithms and the effectiveness of our ranking and filtering semantics. The experimental results show that our approach produces results of higher quality compared to previous ones and our algorithms are fast and scale well with respect to the input and output size.

Efficient keyword search on large tree structured datasets

Keyword search is the most popular paradigm for querying XML data on the web. In this context, three challenging problems are (a) to avoid missing useful results in the answer set, (b) to rank the results with respect to some relevance criterion and (c) to design algorithms that can efficiently compute the results on large datasets. In this paper, we present a novel multi-stack based algorithm that returns as an answer to a keyword query all the results ranked on their size. Our algorithm exploits a lattice of stacks each corresponding to a partition of the keyword set of the query. This feature empowers a linear time performance on the size of the input data for a given number of query keywords. As a result, our algorithm can run efficiently on large input data for several keywords. We also present a variation of our algorithm which accounts for infrequent keywords in the query and show that it can significantly improve the execution time. An extensive experimental evaluation of our approach confirms the theoretical analysis, and shows that it scales smoothly when the size of the input data and the number of input keywords increases.

Keyword Pattern Graph Relaxation for Selective Result Space Expansion on Linked Data

Keyword search is a popular technique for querying the ever growing repositories of RDF graph data. In recent years different approaches leverage a structural summary of the graph data to address the typical keyword search related problems. These approaches compute queries pattern graphs corresponding to alternative interpretations of the keyword query and the user selects one that matches her intention to be evaluated against the data. Though promising, these approaches suffer from a drawback: because summaries are approximate representations of the data, they might return empty answers or miss results which are relevant to the user intent. In this paper, we present a novel approach which combines the use of the structural summary and the user feedback with a relaxation technique for pattern graphs. We leverage pattern graph homomorphisms to define relaxed pattern graphs that are able to extract more results potentially of interest to the user. We introduce an operation on pattern graphs and we show that it can produce all relaxed pattern graphs. To guarantee that the result pattern graphs are as close to the initial pattern graph as possible, we devise different metrics to measure the degree of relaxation of a pattern graph. We design an algorithm that computes relaxed pattern graphs with non-empty answers in relaxation order. Finally, we run experiments to measure the effectiveness of our ranking of relaxed pattern graphs and the efficiency of our system.

Diversifying the Results of Keyword Queries on Linked Data

Keyword search is a popular technique for retrieving information from the ever growing repositories of RDF graph data on the Web. However, keyword queries are inherently ambiguous, resulting in an overwhelming number of candidate results. These results correspond to different interpretations of the query. Most of the current keyword search approaches ignore the diversity of the result interpretations and might fail to provide a broad overview of the query aspects to the users who are interested in exploratory search. To address this issue, we introduce in this paper, a novel technique for diversifying keyword search results on RDF graph data. We generate pattern graphs which are structured queries corresponding to alternative interpretations of the given keyword query. We model the problem as an optimization problem aiming at selecting a set of k pattern graphs with maximum diversity. We devise a metric to estimate the diversity of a set of pattern graphs, and we design an algorithm that employs a greedy heuristic to generate a diverse list of k pattern graphs for a given keyword query.

Incorporating Cohesiveness into Keyword Search on Linked Data

Keyword search is a popular technique for querying the ever increasing repositories of RDF graph data because it frees the user from knowing a formal query language and the structure of the data. However, the imprecision of keyword queries results in overwhelming numbers of candidate results making the identification of relevant results challenging and hindering the scalability of the query evaluation algorithms.

Application of the peer-to-peer paradigm in digital libraries

We present the architecture of a largely distributed Digital Library that is based on the Peer-to-Peer computing paradigm. The three goals of the architecture are: (i) increased node autonomy, (ii) flexible location of data, and (iii) efficient query evaluation. To satisfy these goals we propose a solution based on schema mappings and query reformulation. We identify the problems involved in developing a system based on the proposed architecture and present ways of tackling them. A prototype implementation provides encouraging results.