No abstract

Data mining is the discovery of interesting and hidden patterns from a large amount of collected data. Applications can be found in many organisations with large databases, for many different purposes such as customer relationships, marketing, planning, scientific discovery, and other data analysis. In this paper, the problem of mining N-most interesting itemsets is addressed. We make use of the techniques of COFI-tree in order to tackle the problem. In our experiments, we find that our proposed algorithm based on COFI-tree performs faster than the previous approach BOMO based on the FP-tree.

Stimulated by the outstanding membrane properties of poly(amidesulfonamide)s (PASAs), 2 series of copolymers were synthesized by low-temperature solution polycondensation of either the diamino monomer N, or N,NЈ-bis(4-aminophenylsulfonyl)-2,5-dimethylpiperazine (1B) with a mixture of isophthaloyl and terephthaloyl chloride. All of these copolymers were adequately characterized by intrinsic viscosity measurements, infrared (IR) spectroscopy, 1 H and 13 C nuclear magnetic resonance, differential scanning calorimetry, thermogravimetric analysis, and wide-angle X-ray diffraction. The effects of various feed ratios of isophthaloyl-terephthaloyl chloride on the physical properties of the copolymers were investigated. In particular, the crystallinity of PASA copolymers was found to be trimmed down by copolymerizing diamino monomer 1A or 1B with increasing fraction of isophthaloyl chloride. It was envisioned that when the crystallinity of the PASA membranes decreased, permeation flux would be preferably increased at a minimum sacrifice of separation characteristics in the pervaporation process. Experimental findings, however, revealed that the resulting permeation flux and separation factor of pervaporation membranes derived from various copolymers were both within the same order of magnitude.