Search

With the rapid development of distributed transactional databases in recent years, there is an urgent need for fair performance evaluation and comparison. Though there are various open-source benchmarks built for databases, it is lack of a comprehensive study about the applicability for distributed transactional databases. This paper presents a review of the state-of-art benchmarks with respect to distributed transactional databases. We first summarize the representative architectures of distributed transactional databases and then provide an overview about the chock points in distributed transactional databases. Then, we classify the classic transactional benchmarks based on their characteristics and design purposes. Finally, we review these benchmarks from schema and data definition, workload generation, and evaluation and metrics to check whether they are still applicable to distributed transactional databases with respect to the chock points. This paper exposes a potential research direction to motivate future benchmark designs in the area of distributed transactional databases.

Idiomaticity refers to the situation where the meaning of a lexical unit cannot be derived from the usual meanings of its constituents. As a ubiquitous phenomenon in languages, the existence of idioms often causes significant challenges for semantic NLP tasks. While previous research mostly focuses on the idiomatic usage detection of English verb-noun combinations and the semantic analysis of Noun Compounds (NCs), the idiomaticity issues of Chinese NCs have been rarely studied. In this work, we aim at classifying Chinese NCs into four idiomaticity degrees. Each idiomaticity degree refers to a specific paradigm of how the NCs should be interpreted. To address this task, a Relational and Compositional Representation Learning model (RCRL) is proposed, which considers the relational textual patterns and the compositionality levels of Chinese NCs. RCRL learns relational representations of NCs to capture the semantic relations between two nouns within an NC, expressed by textual patterns and their statistical signals in the corpus. It further employs compositional representations to model the compositionality levels of NCs via network embeddings. Both loss functions of idiomaticity degree classification and representation learning are jointly optimized in an integrated neural network. Experiments over two datasets illustrate the effectiveness of RCRL, outperforming state-of-the-art approaches. Three applicational studies are further conducted to show the usefulness of RCRL and the roles of idiomaticity prediction of Chinese NCs in the fields of NLP.

Unbounded,unordered and large scale datasets are increasingly common in recent years.Meanwhile,the processing requirements from data consumers are becoming more and more sophisticated,such as event time,window and latency.In order to deal with the evolved processing requirements on these unbounded,unordered and large scale datasets,the dataflow model in big data processing was introduced.On one hand,the dataflow graph of the dataflow model in big data processing was analyzed from the level of execution engine.On other hand,the dataflow programming model of the dataflow model in big data processing was analyzed from the level of unified programming.Furthermore,the different implementations of dataflow graph and dataflow programming model in multiple execution engines were analyzed,including Spark,a batch processing engine,and Flink,a stream processing engine.

Database benchmarking has stimulated the development of data management systems and technologies. In big data environments, benchmarking should be revisited. Therefore, research on benchmarks for big data management systems is a key problem for big data research and applications. Benchmark design can be achieved from three different perspectives, i.e. data, workload, and performance measurements. After the brief introduction to these three elements and the progress of big data management system research, the requirements and challenges to benchmarking big data management systems were analyzed. Through the introduction to a benchmark for analytical queries over social media data, named as BSMA, the issues of design and implementation of a benchmark for big data management systems were discussed.

The rapid development of the financial technology has put forward great challenge to the database systems in bank. Specific system features, such as high performance, scalability, high availability and fault tolerance, were demanded. The traditional database systems are difficult to meet these requirements at the same time. In response to these challenges and meet the need of government, which calls for the “independent control of those core technology”, the Bank of Communications focuses on developing a new generation of database systems that support mission-critical banking transactions. A distributed database system with high performance, scalability, high availability and fault-tolerant characteristics was implemented by developing the lightweight distributed election protocol and distributed transaction, and it was applied to a number of important applications.

Blockchain could achieve trustworthy ledger management without central node.It has been successfully supporting financial applications such as Bitcoin.The essence of blockchain is trusted data management in an incompletely trusted environment.It has such features as decentralization,anti-tampering,strong consistency and integrity.On the other hand,blockchain also suffers the problem of weak data management functions,low performance and other issues.By comparing blockchain with traditional data management techniques and analyzing the three applications of blockchain beyond the financial fields,new research issues on blockchain were explored.The requirement for design and implementation of sharing database systems for specific application domains was discussed.The sharing database can be simply defined as a database that supports mission-critical applications,the sharing economic business model,and even itself is implemented in a sharing economy manner.

Big data is still a buzzword，and more and more people are talking about it with various kinds of different explanations.Based on writer’s understanding，the big data，big data strategy and “internet plus” initiative will be discussed here.The database philosophy was revisited，for understanding the development of data management is meaningful to catch the good opportunities in big data era.Moreover，from the point of view of a senior IT professional，the development paradigm for IT has been shifted in the past decade.The change was described，and three systems and their development and deployment were presented.A new concept，sharing database，was proposed to catch up the notion behind the block chain.

Recently,graph data has been widely used in domains such as information security,scientific research,internet services,etc.,that stimulates the fast development of graph data management systems.However,existing benchmarks for graph databases are all designed for applications that manage and analyze social networks.The statistical characteristics of knowledge graphs were analyzed,and compared with two social networks.It was showed that knowledge graphs,as an important and fast growing kind of graph data,were significantly different from social networks.Therefore,existing social network based benchmarks were not suitable for applications that deal with knowledge graphs.Furthermore,the requirements for a new benchmark were analyzed.

There are some characteristics for IT development in the big data era:the real-life applications are the driving force for innovation; open sourcing accelerates innovation,and the advancement in hardware lay the foundation for innovation.The data sciences and engineering was regarded as an emerging and developing interdisciplinary and discussed from the aspects such as social innovation and development,talents demand changes,and technology development.Then the features,connotations,and knowledge hierarchy of data sciences and engineering as a discipline were described.Finally,the associated research and development,talent training,and best practice were also presented.