Contribution towards understanding the categorisation of landforms.

Dissertation submitted in partial fulfilment of the requirements for the Degree of Master of Science in Geospatial Technologies.

Hierarchical categorisation of tags for delicious

In the scenario of social bookmarking, a user browsing the Web bookmarks web pages and assigns free-text labels (i.e., tags) to them according to their personal preferences. In this technical report, we approach one of the practical aspects when it comes to represent users' interests from their tagging activity, namely the categorization of tags into high-level categories of interest. The reason is that the representation of user profiles on the basis of the myriad of tags available on the Web is certainly unfeasible from various practical perspectives; mainly concerning the unavailability of data to reliably, accurately measure interests across such fine-grained categorisation, and, should the data be available, its overwhelming computational intractability. Motivated by this, our study presents the results of a categorization process whereby a collection of tags posted at Delicious #http://delicious.com# are classified into 200 subcategories of interest.

Bag-of-Features Approach to Unsupervised Visual Object Categorisation

The large and growing number of digital images is making manual image search laborious. Only a fraction of the images contain metadata that can be used to search for a particular type of image. Thus, the main research question of this thesis is whether it is possible to learn visual object categories directly from images. Computers process images as long lists of pixels that do not have a clear connection to high-level semantics which could be used in the image search. There are various methods introduced in the literature to extract low-level image features and also approaches to connect these low-level features with high-level semantics. One of these approaches is called Bag-of-Features which is studied in the thesis. In the Bag-of-Features approach, the images are described using a visual codebook. The codebook is built from the descriptions of the image patches using clustering. The images are described by matching descriptions of image patches with the visual codebook and computing the number of matches for each code. In this thesis, unsupervised visual object categorisation using the Bag-of-Features approach is studied. The goal is to find groups of similar images, e.g., images that contain an object from the same category. The standard Bag-of-Features approach is improved by using spatial information and visual saliency. It was found that the performance of the visual object categorisation can be improved by using spatial information of local features to verify the matches. However, this process is computationally heavy, and thus, the number of images must be limited in the spatial matching, for example, by using the Bag-of-Features method as in this study. Different approaches for saliency detection are studied and a new method based on the Hessian-Affine local feature detector is proposed. The new method achieves comparable results with current state-of-the-art. The visual object categorisation performance was improved by using foreground segmentation based on saliency information, especially when the background could be considered as clutter.

Determination of the silvo-melliferous regions of Benin: a nationwide categorisation of the land based on melliferous plants suitable for timber production

Perennial plants are the main pollen and nectar sources for bees in the tropical areas where most of the annual flora are burned in dry seasons. Therefore perennial plants constitute the most reliable bio materials for determining and evaluating the beekeeping regions of the Republic of Benin. A silvo-melliferous region (S-MR) is a geographical area characterised by a particular set of homogenous melliferous plants that can produce timber. Using both the prevailing climatic and the agro-ecological conditions six S-MRs could be identified, i.e. the South region, the Common Central region, the Central West region, the Central North region, the Middle North region and the Extreme North region. At the country level, the melliferous plants were dominated by Vitellaria paradoxa which is common to all regions. The most diversified family was the Caesalpiniaceae (12 species) followed by the Combretaceae (10 species) and Combretum being the richest genus. The effect of dominance is particularly high in the South region where Elaeis guineensis alone represented 72.6% of the tree density and 140% of the total plant importance. The total melliferous plant density varied from 99.3 plants ha^(−1) in the Common Central region to 178.0 plants ha^(−1) in the Central West region. On the basis of nectar and pollen source, the best region for beekeeping is the CentralWest region with 46.7% of nectar producing trees, 9.4% of pollen producing trees and 40.6% of plants that issue both, this in opposition to the South region which was characterised by an unbalanced distribution of melliferous trees.

Categorisation of synoptic environments associated with mesoscale convective systems over the UK

Mesoscale convective systems (MCSs) are relatively rare events in the UK but, when they do occur, can be associated with weather that is considered extreme with respect to climatology (as indicated by the number of such events that have been analysed as case studies). These case studies usually associate UK MCSs with a synoptic environment known as the Spanish plume. Here a previously published 17 year climatology of UK MCS events is extended to the present day (from 1998 to 2008) and these events classified according to the synoptic environment in which they form. Three distinct synoptic environments have been identified, here termed the classical Spanish plume, modified Spanish plume, and European easterly plume. Detailed case studies of the two latter, newly defined, environments are presented. Composites produced for each environment further reveal the differences between them. The classical Spanish plume is associated with an eastward propagating baroclinic cyclone that evolves according to idealised life cycle 1. Conditional instability is released from a warm moist plume of air advected northeastwards from Iberia that is capped by warmer, but very dry air, from the Spanish plateau. The modified Spanish plume is associated with a slowly moving mature frontal system associated with a forward tilting trough (and possibly cut-off low) at 500 hPa that evolves according to idealised life cycle 2. As in the classical Spanish plume, conditional instability is released from a warm plume of air advected northwards from Iberia. The less frequent European easterly plume is associated with an omega block centred over Scandinavia at upper levels. Conditional instability is released from a warm plume of air advected westwards across northern continental Europe. Unlike the Spanish plume environments, the European easterly plume is not a warm sector phenomena associated with a baroclinic cyclone. However, in all environments the organisation of convection is associated with the interaction of an upper-level disturbance with a low-level region of warm advection.

Keyphrase extraction by synonym analysis of n-grams for e-journals categorisation

Automatic keyword or keyphrase extraction is concerned with assigning keyphrases to documents based on words from within the document. Previous studies have shown that in a significant number of cases author-supplied keywords are not appropriate for the document to which they are attached. This can either be because they represent what the author believes the paper is about not what it actually is, or because they include keyphrases which are more classificatory than explanatory e.g., “University of Poppleton” instead of “Knowledge Discovery in Databases”. Thus, there is a need for a system that can generate appropriate and diverse range of keyphrases that reflect the document. This paper proposes a solution that examines the synonyms of words and phrases in the document to find the underlying themes, and presents these as appropriate keyphrases. The primary method explores taking n-grams of the source document phrases, and examining the synonyms of these, while the secondary considers grouping outputs by their synonyms. The experiments undertaken show the primary method produces good results and that the secondary method produces both good results and potential for future work.

Automated categorisation of e-journals by synonym analysis of n-grams

Automatic keyword or keyphrase extraction is concerned with assigning keyphrases to documents based on words from within the document. Previous studies have shown that in a significant number of cases author-supplied keywords are not appropriate for the document to which they are attached. This can either be because they represent what the author believes a paper is about not what it actually is, or because they include keyphrases which are more classificatory than explanatory e.g., “University of Poppleton” instead of “Knowledge Discovery in Databases”. Thus, there is a need for a system that can generate an appropriate and diverse range of keyphrases that reflect the document. This paper proposes two possible solutions that examine the synonyms of words and phrases in the document to find the underlying themes, and presents these as appropriate keyphrases. Using three different freely available thesauri, the work undertaken examines two different methods of producing keywords and compares the outcomes across multiple strands in the timeline. The primary method explores taking n-grams of the source document phrases, and examining the synonyms of these, while the secondary considers grouping outputs by their synonyms. The experiments undertaken show the primary method produces good results and that the secondary method produces both good results and potential for future work. In addition, the different qualities of the thesauri are examined and it is concluded that the more entries in a thesaurus, the better it is likely to perform. The age of the thesaurus or the size of each entry does not correlate to performance.

Unconscious effects of grammatical gender during object categorisation

Does language modulate perception and categorisation of everyday objects? Here, we approach this question from the perspective of grammatical gender in bilinguals. We tested Spanish–English bilinguals and control native speakers of English in a semantic categorisation task on triplets of pictures in an all-in-English context while measuring event-related brain potentials (ERPs). Participants were asked to press a button when the third picture of a triplet belonged to the same semantic category as the first two, and another button when it belonged to a different category. Unbeknownst to them, in half of the trials, the gender of the third picture name in Spanish had the same gender as that of the first two, and the opposite gender in the other half. We found no priming in behavioural results of either semantic relatedness or gender consistency. In contrast, ERPs revealed not only the expected semantic priming effect in both groups, but also a negative modulation by gender inconsistency in Spanish–English bilinguals, exclusively. These results provide evidence for spontaneous and unconscious access to grammatical gender in participants functioning in a context requiring no access to such information, thereby providing support for linguistic relativity effects in the grammatical domain.

Universal and language-specific patterns of categorisation

Previous work on object classification preferences shows that speakers of languages that lack morphological plural marking (like Yucatec and Japanese) display a tendency to match objects by common material, while speakers of languages with morphological plural marking (like English) display a tendency to match objects by common shape. The present paper compares categorisation preferences of English and Japanese speakers with those of Greek speakers. Greek resembles English in that it has morphological plural marking, but contrasts with English in that mass nouns typically do not resist pluralization. Results show that all groups distinguish significantly between countable objects and non-countable substances, but the degree to which they do this differs and conforms to language-specific grammatical patterns. It is argued that the effects of grammatical structure on categorisation preferences are finer-grained than earlier studies have assumed, thus providing a more precise account of the extent and nature of linguistic influence on cognition.

The Role of Multiple Social Categorisation in Promoting the Inclusion in the Human Group of Outgroup Members

Training selective use of a complex categorisation strategy

The present study aims to encourage selective use of a complex categorisation strategy. More specifically, participants will be trained to use a two dimensional strategy in one region of category space and a more complex three-dimensional strategy in another region of category space. In the 2–3 conditions, participants will be presented with stimuli requiring the two-dimensional strategy in the first phase of training and the three-dimensional strategy in the second phase of training. In the 3-2 conditions, participants will be presented with stimuli requiring the three-dimensional strategy in the first phase of training and the two-dimensional strategy in the second phase of training. The main dependent measure will be performance on exceptions to the two-dimensional strategy. If participants learn to selectively use the three-dimensional strategy, then we expect them to correctly classify novel exceptions that occur in the three-dimensional region of the category space and incorrectly classify novel exceptions that occur in the two-dimensional region of the category space.