Bonan Min

Infusion of Labeled Data into Distant Supervision for Relation Extraction

Distant supervision usually utilizes only unlabeled data and existing knowledge bases to learn relation extraction models. However, in some cases a small amount of human labeled data is available. In this paper, we demonstrate how a state-of-theart multi-instance multi-label model can be modified to make use of these reliable sentence-level labels in addition to the relation-level distant supervision from a database. Experiments show that our approach achieves a statistically significant increase of 13.5% in F-score and 37% in area under the precision recall curve.

i-DBF: an Improved Bloom Filter Representation Method on Dynamic Set

Bloom filter is a simple space-efficient randomized data structure for representing a set in order to support membership queries, which uses an m-bit array to represent a data set. Dynamic bloom filter (DBF) can support concisely representation and approximate membership queries of dynamic set instead of static set. It has been proved that DBF not only possess the advantage of standard bloom filter, but also has better features when dealing with dynamic set. But DBF also has a disadvantage: the addition operation which mapped element x into bloom filter s will become no sense, if some of the first s-1 bloom filters have already responded that element x is in set A with some false positive probability. We point out this shortcoming and improve the addition operation with a new algorithm. We call this improved dynamic bloom filter i-DBF. Finally, we prove that this i-DBF has better performance both in the storage space and in the false positive probability

SODA: Towards a framework for self optimization via demand adaptation in Peer-to-Peer networks

Peer-to-Peer (P2P) applications have been consuming an increasingly significant fraction of Internet bandwidth. They are becoming a financial burden to Internet Service Providers (ISPs), creating hot spots in the Internet, and causing potential performance degradation to other applications. As a result, there has been increasing tensions between P2P applications and network service providers. In this paper, we propose a framework called SODA for P2P applications to be self-adaptive and optimize their demands in order to more efficiently utilize network resources. Through preliminary experiments using two representative P2P applications, we demonstrate that SODA can effectively reduce bandwidth consumption by adapting the demands among peers.

A simple, universal and scalable approach to migrate applications to hybrid network using ALG

This paper presents a universal approach using application layer gateway (ALG) mechanism to enable applications supporting both IPv4 and IPv6 network, which is simple and easy to implement as compared to solutions on networking layer. Maze is a P2P file sharing system, which is developed, deployed and operated by our academic research team. Using this approach, we improved Maze system to a new version which is called next generation Maze (ngMaze) to support the hybrid network environment. In this paper, first we present the detail design and implementation of ngMaze system architecture. Then, we evaluate our method and give some conclusions. In the final, we give an advanced approach and describe our future work