In keyword search over a data graph, an answer is a non-redundant subtree that contains all the keywords of the query. A naive approach to producing all the answers by increasing height is to generalize Dijkstra's algorithm to enumerating all acyclic paths by increasing weight. The idea of freezing is introduced so that (most) non-shortest paths are generated only if they are actually needed for producing answers. The resulting algorithm for generating subtrees, called GTF, is subtle and its proof of correctness is intricate. Extensive experiments show that GTF outperforms existing systems, even ones that for efficiency's sake are incomplete (i.e., cannot produce all the answers). In particular, GTF is scalable and performs well even on large data graphs and when many answers are neede

A system for keyword search on data graphs is demonstrated on two challenging datasets: the large DBLP and Mondial (which is highly cyclic and has a complex schema). The system supports search, exploration and question answering. The demonstration shows how the system copes with the main challenges in keywords search on data graphs. In particular, the system generates answers efficiently and completely (i.e., it does not miss answers). It has an effective ranking mechanism that also takes into account redundancies among answers. Finally, the system uses a novel technique for displaying multi-node subtrees in a compact graphical form that facilitates quick and easy understanding, which is essential for effective browsing of the answers.

In keyword search over data graphs, an answer is a nonredundant subtree that includes the given keywords. An algorithm for enumerating answers is presented within an architecture that has two main components: an engine that generates a set of candidate answers and a ranker that evaluates their score. To be effective, the engine must have three fundamental properties. It should not miss relevant answers, has to be efficient and must generate the answers in an order that is highly correlated with the desired ranking. It is shown that none of the existing systems has implemented an engine that has all of these properties. In contrast, this paper presents an engine that generates all the answers with provable guarantees. Experiments show that the engine performs well in practice. It is also shown how to adapt this engine to queries under the OR semantics. In addition, this paper presents a novel approach for implementing rankers destined for eliminating redundancy. Essentially, an answer is ranked according to its individual properties (relevancy) and its intersection with the answers that have already been presented to the user. Within this approach, experiments with specific rankers are described.