Pavel Praks

Information retrieval in hydrochemical data using the latent semantic indexing approach

Kwok-wing Chau (corresponding author) Nitin Muttil Department of Civil and Structural Engineering, Hong Kong Polytechnic University, Hung Hom, Hong Kong, China Tel: +852 2766 6014 E-mail: cekwchau@polyu.edu.hk In this study, data mining using box plots and multivariate statistical analysis using factor analysis are employed for a spatio-temporal analysis of coastal water quality data from Tolo Harbour, Hong Kong. The analysis of box plots reveals pronounced spatial heterogeneity of the parameters studied. The spatial analysis clearly shows monitoring station TM2 in the Harbour Subzone to be most susceptible to eutrophication with the highest nutrient and algal biomass concentrations. The factor analysis brings to light dominant parameters to the ecological system under the coastal marine environment. The temporal analysis confirms the considerable decline in nutrient levels in recent years. In spite of this decline, the factor analysis indicates that nutrient processes play an important role even in recent years, suggesting an adequate supply of nutrients. It seems that they are being released from sources other than known point sources, possibly from nutrients accumulated in the sediments, necessitating steps to be undertaken for their control also. This study demonstrates the use of data mining techniques in the ecological system in Tolo Harbour.

Information Extraction from HTML Product Catalogues: From Source Code and Images to RDF

We describe an application of information extraction from company Web sites focusing on product offers. A statistical approach to text analysis is used in conjunction with different ways of image classification. Ontological knowledge is used to group the extracted items into structured objects. The results are stored in an RDF repository and made available for structured search.

On SVD-free latent semantic indexing for image retrieval for application in a hard industrial environment

Retrieval system based on latent semantic indexing (LSI) is presented in our paper. LSI is usually based on single value decomposition, which is time-consuming computation. New, fast algorithm for computing the LSI was developed. The algorithm was tested on text and image data collection. Experimental results are given. The algorithms are intended for practical industrial use within a control and information system operating at coking plant at Nova hut Ostrava. The description of the industrial environment is given.

On SVD-Free Latent Semantic Indexing for Iris Recognition of Large Databases

This chapter presents a method for an automatic identification of persons by iris recognition. A raster image of a human iris is represented as a sequence of color pixels. Information retrieval is conducted by the Latent Semantic Indexing (LSI) method. The pattern recognition algorithm is powered very effectively when the time-consuming Singular Value Decomposition (SVD) of LSI is replaced by the partial symmetric eigenproblem. Numerical experiments on a real 488 MB biometric data collection indicates feasibility of the presented approach as a tool for automated image recognition without special preprocessing.

Measurement placement algorithms for state estimation when the number of phasor measurements by each PMU is limited

It is essential for Energy Management Systems (EMS) applications and control centre technologies that the measurements allow monitoring of the power system. An important tool for the EMS is state estimation which makes it possible to estimate the state variables of the power system based on these measurements. The measurements should allow observability of the power system which means they can be used to describe the system. The objective of the paper is to develop an algorithm for placement of PMUs in order to make the system observable. The novelty is that a limit on the number of measurements each PMU can make is introduced. The number of PMUs needed to make the network observable will therefore depend on the maximum number of measurements each PMU can make. The algorithm is tested on three IEEE test networks.

Human expert modelling using semantics-oriented video retrieval for surveillance in hard industry

The surveillance application aims at improving the quality of technology via modelling human expert behaviour in the coking plant Mittal Steel Ostrava, the Czech Republic. Video data on several industrial processes are captured by means of a CCD camera and classified by using Latent Semantic Indexing (LSI) with the respect to etalons classified by an expert. We also investigate the possibility to combine LSI with color histogram analysis, using experience from our prior research on web image classification.

Statistical model of quality of radon measurements using electret ion chamber detectors

This paper presents a statistical model for estimating probability, which states that a quality of radon measurements using the electret ion chamber system RM-1 will be accurate. The quality of the electret measurement was modelled as a ratio between the reference radon concentration and values measured under real conditions with varied levels of indoor radon concentration and microclimate indoor conditions. It was stated that the accuracy tolerance of measured values is 20 % of the given reference value. To estimate the uncertainty of the statistical model, the exact confidence limits for the estimated probabilities are computed. The statistical model was confirmed by an independent set of measurements. Moreover, the effect of absolute humidity on the quality estimation of electret detectors are also statistically analysed and discussed. The results of the statistical model confirm that the electret system is robust and suitable for estimation of radon concentration.

Monitoring the impact of the intensity of blowing of an inert gas to the visual character of the molten steel surface

Steel Companies use a ladle furnace refining process. The main feature of this ladle technology exists in refining under non-oxidizing atmosphere and support of slag-metal reaction through stirring by Ar gas injection, for a desired long time due to the temperature compensation by arc heating. A CCD camera, placed above the ladle monitors the slag formation (so called eyes) that has a different temperature and therefore a different color than the metal. We developed the automatic software capable to analyze the homogeneity of the surface and characterize features of the molten steel level such us presence of slag clusters. In our paper, the method for visual monitoring and automated retrieval of slag clusters taken from the CCD camera will be described.

Automatic Object Classification and Image Retrieval by Sobel Edge Detection and Latent Semantic Methods

We perform in this paper a comparative study of ability of the proposed novel image retrieval algorithms to provide automated object classification invariant of rotation, translation and scaling. We analyze simple cosine similarity coefficient methods and the SVD-free Latent Semantic method with an alternative sparse representation of color images. Considering applied cosine similarity coefficient methods, the two following approaches were tested and compared: i) the processing of the whole image and ii) the processing of the image that contains edges extracted by the application of the Sobel edge detector. Numerical experiments on a real database sets indicate feasibility of the presented approach as automated object classification tool without special image pre-processing.

Linear algebra for vision-based surveillance in heavy industry - convergence behavior case study

The surveillance application aims at improving the quality of technology via modelling human expert behaviour in the coking plant ArcelorMittal Ostrava, the Czech Republic. Video data on several industrial processes are captured by means of a CCD camera and classified by using Latent Semantic Indexing (LSI) with the respect to etalons classified by an expert. We also study the convergence behavior of proposed partial eigenproblem-based dimension reduction technique and its ability for knowledge acquisition. Having increased the computational effort of the dimension reduction technique did not imply the increasing quality of retrieved results in our cases.