Keyword search in graphs, relational databases and social networks

Keyword search, a well known mechanism for retrieving relevant information from a set of documents, has recently been studied for extracting information from structured data (e.g., relational databases and XML documents). It offers an alternative way to query languages (e.g., SQL) to explore databases, which is effective for lay users who may not be familiar with the database schema or the query language. This dissertation addresses some issues in keyword search in structured data. Namely, novel solutions to existing problems in keyword search in graphs or relational databases are proposed. In addition, a problem related to graph keyword search, team formation in social networks, is studied. The dissertation consists of four parts.

The first part addresses keyword search over a graph which finds a substructure of the graph containing all or some of the query keywords. Current methods for keyword search over graphs may produce answers in which some content nodes (i.e., nodes that contain input keywords) are not very close to each other. In addition, current methods explore both content and non-content nodes while searching for the result and are thus both time and memory consuming for large graphs. To address the above problems, we propose algorithms for finding r-cliques in graphs. An r-clique is a group of content nodes that cover all the input keywords and the distance between each pair of nodes is less than or equal to r. Two approximation algorithms that produce r-cliques with a bounded approximation ratio in polynomial delay are proposed.

In the second part, the problem of duplication-free and minimal keyword search in graphs is studied. Current methods for keyword search in graphs may produce duplicate answers that contain the same set of content nodes. In addition, an answer found by these methods may not be minimal in the sense that some of the nodes in the answer may contain query keywords that are all covered by other nodes in the answer. Removing these nodes does not change the coverage of the answer but can make the answer more compact. We define the problem of finding duplication-free and minimal answers, and propose algorithms for finding such answers efficiently.

Meaningful keyword search in relational databases is the subject of the third part of this dissertation. Keyword search over relational databases returns a join tree spanning tuples containing the query keywords. As many answers of varying quality can be found, and the user is often only interested in seeing the·top-k answers, how to gauge the relevance of answers to rank them is of paramount importance. This becomes more pertinent for databases with large and complex schemas. We focus on the relevance of join trees as the fundamental means to rank the answers. We devise means to measure relevance of relations and foreign keys in the schema over the information content of the database.

The problem of keyword search over graph data is similar to the problem of team formation in social networks. In this setting, keywords represent skills and the nodes in a graph represent the experts that possess skills. Given an expert network, in which a node represents an expert that has a cost for using the expert service and an edge represents the communication cost between the two corresponding experts, we tackle the problem of finding a team of experts that covers a set of required skills and also minimizes the communication cost as well as the personnel cost of the team. We propose two types of approximation algorithms to solve this bi-criteria problem in the fourth part of this dissertation.