Umeå universitets logga

This paper introduces an automatic algorithm that can be seen as an extension to the Eclat algorithm, as well as a corresponding proof of concept prototype. It uses inverted indices and statistical pruning of the possible solution space as early as possible.

We describe a way to use peer-graded oral presentations as a way of reducing the load on the teacher, and show that almost identical results as can be achieved as with teacher graded presentations. Moreover, we have found that very little in the form of explicit criteria are needed.

In this paper we present a free dataset, usable for testing web search engines.  The dataset corresponds to a snapshot of the Nordic part of the Internet in early 2007 and is highly abstracted, with numbers representing each web page.  The released dataset consists of three parts; a graph, 76 sets of pages containing each tested word combination, and some files to use when calculating relevance of the resulting sets of algorithms/search engines. We also present a new compound statistic as well as statistical results for some search engine and information retrieval algorithms.

In this paper we present a free dataset, usable for testing web search engines. The dataset corresponds to a snapshot of the Nordic part of the Internet back in early 2007 and is highly abstracted, with numbers representing each web page. The released dataset consists of three parts; a graph, 76 sets of pages containing each tested word combination, and some files to use when calculating relevance of the resulting sets of algorithms/search engines. We also present statistics for some search engine algorithms.

Purpose – The purpose of this paper is to assign topic-specific ratings to web pages.

Design/methodology/approach – The paper uses power iteration to assign topic-specific rating values (called relevance) to web pages, creating a ranking or partial order among these pages for each topic. This approach depends on a set of pages that are initially assumed to be relevant for a specific topic; the spatial link structure of the web pages; and a net-specific decay factor designated ξ.

Findings – The paper finds that this approach exhibits desirable properties such as fast convergence, stability and yields relevant answer sets. The first property will be shown using theoretical proofs, while the others are evaluated through stability experiments and assessments of real world data in comparison with already established algorithms.

Research limitations/implications – In the assessment, all pages that a web spider was able to find in the Nordic countries were used. It is also important to note that entities that use domains outside the Nordic countries (e.g..com or.org) are not present in the paper's datasets even though they reside logically within one or more of the Nordic countries. This is quite a large dataset, but still small in comparison with the entire worldwide web. Moreover, the execution speed of some of the algorithms unfortunately prohibited the use of a large test dataset in the stability tests.

Practical implications – It is not only possible, but also reasonable, to perform ranking of web pages without using Markov chain approaches. This means that the work of generating answer sets for complex questions could (at least in theory) be divided into smaller parts that are later summed up to give the final answer.

Originality/value – This paper contributes to the research on internet search engines.

A comparison between a number of search engines from three different families (HITS, PageRank, and Propagation of Trust) is presented for a small web server with respect to perceived relevance. A total of 307 individual tests have been done and the results from these were disseminated to the algorithms, and then handled using confidence intervals, Kolmogorov-Smirnov and ANOVA. We show that the results can be grouped according to algorithm family, and also that the algorithms (or at least families) can be partially ordered in order of relevance.

In this paper we propose an algorithm that automatically creates concept hierarchies or lattices for discrete databases and datasets. The reason for doing this is to accommodate later data mining operations on the same sets of data without having an expert create these hierarchies by hand.

Each step of the algorithm will be examined; We will show inputs and output for each step using a small example. The theoretical upper bound of the complexity for each part of the algorithm will be presented, as well as real time measurements for a number of databases. We will finally present a time model of the algorithm in terms of a number of attributes of the databases

In this paper we propose an algorithm for automatic creation of concept hierarchies from discrete databases and datasets. The reason for doing this is to accommodate later data mining operations on the same set of data without having an expert create these hierachies by hand.

We will go through the algorithm thoroughly and show the results from each step of the algorithm using a (small) example. We will also give actual execution times for our prototype for non-trivial example data sets and estimates of the complexity of the algorithm in terms of the number of records and the number of distinct data values in the data set.

ALGEXT is a program that extracts strategic/block comments from C source files to improve maintainability and to keep documentation consistent with source code. This is done by writing the comments in the source code in what we call extractable algorithms, describing the algorithm used in the functions.

ALGEXT recognizes different kinds of comments:

• Strategic comments are comments that proceed a block of code, with only whitespace preceding it on the line,
• Tactical comments are comments that describes the code that precedes it on the same line,
• Function comments are comments immediately preceding a function definition, describing the function,
• File comments are comments at the head of the file, before any declarations of functions and variables, and finally
• Global comments are comments within the global scope, but not associated with a function.

Only strategic comment are used as basis for algorithm extraction in ALGEXT.

The paper discusses the rationale for ALGEXT and describes its implementation and usage. Examples are presented for clarification of what can be done with ALGEXT.

Our experience shows that students who use ALGEXT for preparing theirassignments tend to write about 66% more comments than non-ALGEXT users.