Amund Tveit

Peer-to-peer based recommendations for mobile commerce

With the increasing number of mobile commerce facilities, there are challenges in providing customers useful recommendations about interesting products and services. In this paper a Peer-to-Peer (P2P) based collaborative filtering architecture for the support of product and service recommendations for mobile customers is considered. Mobile customers are represented by software assistant agents that act like peers in the processing of recommendations.

Semantic annotation of biomedical literature using google

With the increasing amount of biomedical literature, there is a need for automatic extraction of information to support biomedical researchers. Due to incomplete biomedical information databases, the extraction is not straightforward using dictionaries, and several approaches using contextual rules and machine learning have previously been proposed. Our work is inspired by the previous approaches, but is novel in the sense that it is using Google for semantic annotation of the biomedical words. The semantic annotation accuracy obtained – 52% on words not found in the Brown Corpus, Swiss-Prot or LocusLink (accessed using Gsearch.org) – is justifying further work in this direction.

Multicategory Incremental Proximal Support Vector Classifiers

Support Vector Machines (SVMs) are an efficient data mining approach for classification, clustering and time series analysis. In recent years, a tremendous growth in the amount of data gathered has changed the focus of SVM classifier algorithms from providing accurate results to enabling incremental (and decremental) learning with new data (or unlearning old data) without the need for computationally costly retraining with the old data. In this paper we propose an efficient algorithm for multicategory classification with the incremental proximal SVM introduced by Fung and Mangasarian.

Incremental and decremental proximal support vector classification using decay coefficients

This paper presents an efficient approach for supporting decremental learning for incremental proximal support vector machines (SVM). The presented decremental algorithm based on decay coefficients is compared with an existing window-based decremental algorithm, and is shown to perform at a similar level in accuracy, but providing significantly better computational performance.

Mobile Commerce Agents in WAP-Based Services

With the increasing number of e-commerce services for mobile devices, there are challenges in making these services more personalized and to take into account the severely constrained bandwidth and restricted user interface these devices currently provide. In this paper we consider an agent-based platform for support of mobile commerce using wireless (WAP-based) devices. Agents represent mobile device customers in the network by implementing highly personalized customer profiles. The platform allows customization and adaptation of mobile commerce services as well as pro-active processing and notification of important events. Information to the customers is delivered both via WML-decks and SMS messages. Usage of the platform is illustrated by examples of valued customer membership services and subscription services support. Some details of a prototype platform implementation are briefly considered.

ProtChew: Automatic Extraction of Protein Names from Biomedical Literature

With the increasing amount of biomedical literature, there is a need for automatic extraction of information to support biomedical researchers. Due to incomplete biomedical information databases, the extraction is not straightforward using dictionaries, and several approaches using contextual rules and machine learning have previously been proposed. Our work is inspired by the previous approaches, but is novel in the sense that it is fully automatic and doesn’t rely on expert tagged corpora. The main ideas are 1) unigram tagging of corpora using known protein names for training examples for the protein name extraction classi- fier and 2) tight positive and negative examples by having protein-related words as negative examples and protein names/synonyms as positive examples. We present preliminary results on Medline abstracts about gastrin, further work will be on testing the approach on BioCreative benchmark data sets.

Discovering biological motifs with genetic programming

Choosing the right representation for a problem is important. In this article we introduce a linear genetic programming approach for motif discovery in protein families, and we also present a thorough comparison between our approach and Koza-style genetic programming using ADFs. In a study of 45 protein families, we demonstrate that our algorithm, given equal processing resources and no prior knowledge in shaping of datasets, consistently generates motifs that are of significantly better quality than those we found by using trees as representation. For several of the studied protein families we evolve motifs comparable to those found in Prosite, a manually curated database of protein motifs.Our linear genome gave better results than Koza-style genetic programming for 37 of 45 families. The difference is statistically significant for 24 of the families at the 99% confidence level.

gProt: annotating protein interactions using google and gene ontology

With the increasing amount of biomedical literature, there is a need for automatic extraction of information to support biomedical researchers. Due to incomplete biomedical information databases, the extraction cannot be done straightforward using dictionaries, so several approaches using contextual rules and machine learning have previously been proposed. Our work is inspired by the previous approaches, but is novel in the sense that it combines Google and Gene Ontology for annotating protein interactions. We got promising empirical results – 57.5% terms as valid GO annotations, and 16.9% protein names in the answers provided by our system gProt. The total error-rate was 25.6% consisting mainly of overly general answers and syntactic errors, but also including semantic errors, other biological entities (than proteins and GO-terms) and false information sources.