Today, one of the most important tasks in natural language processing is answering user questions. Especially, users' questions nowadays moved from simple questions to complex questions. In recent years, several question answering datasets have been produced for Persian language, but none of them support complex open-domain and explainable questions. In this article, the PersianMHQA dataset is introduced which is the first open-domain question answering dataset for complex questions based on the unstructured Persian Wikipedia encyclopedia. This dataset contains 7,000 complex questions and sentence-level supporting facts are provided for each question that allows question answering systems to explain the predictions. The questions in this dataset are diverse and explainable and are not limited to any previous knowledge base. Various types of complexity are provided in this dataset, and the questions are designed in such a way that answering them requires reasoning over more than one paragraph.

Keyword search is an alternative for structured languages in querying graph-structured data. A result to a keyword query is a connected structure covering all or part of the queried keywords. The textual coverage and structural compactness have been known as the two main properties of a relevant result to a keyword query. Many previous works examined these properties after retrieving all of the candidate results using a ranking function in a comparative manner. However, this needs a time-consuming search process, which is not appropriate for an interactive system in which the user expects results in the least possible time. This problem has been addressed in recent works by confining the shape of results to examine their coverage and compactness during the search. However, these methods still suffer from the existence of redundant nodes in the retrieved results. In this paper, we introduce the semantic of minimal covered r-clique (MCC r ) for the results of a keyword query as an extended model of existing definitions. We propose some efficient algorithms to detect the MCC r s of a given query. These algorithms can retrieve a comprehensive set of non-duplicate MCC r s in response to a keyword query. In addition, these algorithms can be executed in a distributive manner, which makes them outstanding in the field of keyword search. We also propose the approximate versions of these algorithms to retrieve the top- k approximate MCC r s in a polynomial delay. It is proved that the approximate algorithms can retrieve results in two-approximation. Extensive experiments on two real-world datasets confirm the efficiency and effectiveness of the proposed algorithms.

Unfortunately the second author’s name has been misspelt. It should be read: Abbas NIKNAFS.

Many real-world networks such as Facebook, LinkedIn, and Wikipedia exhibit rich connectivity patterns along with worthwhile content nodes often labeled with meaningful attributes. Keyword search is an effective method to retrieve information from such useful networks. The aim of keyword search is to find a set of answers (subgraphs) covering all or part of the queried keywords. A challenge in keyword search systems is to rank answers according to their relevance to the query. This relevance lies in the textual content and structural compactness of the answers. In this paper, an attribute-specific ranking method is proposed based on language models to rank candidate answers according to their semantic information up to the attribute level. This method scores answers using a model enriched with attribute-specific preferences and integrating both the structure and content of answers. The proposed model is directly estimated on the sub-graphs (answers) and is defined such that it can preserve the local importance of keywords in nodes. Extensive experiments conducted on a standard evaluation framework with three real-world datasets illustrate the superior effectiveness of the proposed ranking method to that of the state-of-the-art methods.

Keyword search (KWS) has been known as an attractive query processor in retrieving information from various types of data which could be modeled as a graph. An answer in response to a keyword query is a set of cohesively connected structures which shows how the data containing query keywords are interconnected in the graph. Finding answers to a given query efficiently and ranking the retrieved answers in an effective way are still two challenging problems in KWS domain. In this paper, we first propose a novel scoring function to optimize the accuracy of ranking the answers. This function is defined based on a carefully designed model called SARM which is an integrated model of the content and structure of an answer. We then develop a two-level KWS approach to support the efficient retrieval of top-k answers to a given query. This approach is based on pruning the search space to concentrate the search on the promising regions. The efficiency of this approach is improved by estimating the boundary scores of the answers in the regions. Extensive experiments conducted on a standard evaluation framework with three real-world datasets confirm the efficiency and effectiveness of the proposed approach.

The ever-increasing volume of trajectories of moving objects and the diversity of intelligent transportation systems and location-based services that rely on spatio-temporal data of moving objects highlight the need for more efficient indexing techniques. The state-of-the-art methods index moving objects at three time modes of past (historical data), present, and future. An integrated method called “PCI” was proposed (past current indexing) to index and store spatial-temporal data of the past and present simultaneously. The method can handle queries in both the time modes and it processes and generates both the past and present indices using an integrated set of processing resources. Two interconnected data structures were utilized to store indices of both the time modes. Connecting the index of different time modes enforces efficiency challenges due to difference in updating costs. Since the method stores the indices in the main memory, the way the structures are connected to each other makes it possible to transfer the current data to the section responsible for historical data. This method indexes them in the trajectory of moving objects at a minimum time expense. As the quality of data inevitably affects the performance of applications, map matching methods was used in PCI to remove noises-e.g., stationary state noises-in the data received from the moving objects. This feature adds to the accuracy and reliability of query results. The effects of data reduction techniques on accelerating indexing, query processing, and reducing memory consumption (in voluminous data sets) were examined. Results of the comparisons, made based on the experiments, showed the higher efficiency of the indexing structure.

Summary These days, large‐scale graph processing becomes more and more important. Pregel, inspired by Bulk Synchronous Parallel, is one of the highly used systems to process large‐scale graph problems. In Pregel, each vertex executes a function and waits for a superstep to communicate its data to other vertices. Superstep is a very time‐consuming operation, used by Pregel, to synchronize distributed computations in a cluster of computers. However, it may become a bottleneck when the number of communications increases in a graph with million vertices. Superstep works like a barrier in Pregel that increases the side effect of skew problem in distributed computing environment. ExPregel is a Pregel‐like model that is designed to reduce the number of communication messages between two vertices resided on two different computational nodes. We have proven that ExPregel reduces the number of exchanged messages as well as the number of supersteps for all graph topologies. Enhancing parallelism in our new computational model is another important feature that manifolds the speed of graph analysis programs. More interestingly, ExPregel uses the same model of programming as Pregel. Our experiments on large‐scale real‐world graphs show that ExPregel can reduce network traffic as well as number of supersteps from 45% to 96%. Runtime speed up in the proposed model varies from 1.2× to 30×. Copyright © 2015 John Wiley & Sons, Ltd.