The influence of pre-processing on the estimation of readability of web documents

This paper investigates the effect that text pre-processing approaches have on the estimation of the readability of web pages. Readability has been highlighted as an important aspect of web search result personalisation in previous work. The most widely used text readability measures rely on surface level characteristics of text, such as the length of words and sentences. We demonstrate that different tools for extracting text from web pages lead to very different estimations of readability. This has an important implication for search engines because search result personalisation strategies that consider users reading ability may fail if incorrect text readability estimations are computed.

Impact of online privacy concerns and brand reputation on consumer willingness to provide personal information

The aim of this research was to identify the role of brand reputation in encouraging consumer willingness to provide personal data online, for the benefits of personalisation. This study extends on Malhotra, Kim and Agarwal’s (2004) Internet Users Information Privacy Concerns Model, and uses the theoretical underpinning of Social Contract Theory to assess how brand reputation moderates the relationship between trusting beliefs and perceived value (Privacy Calculus framework) with willingness to give personal information. The research is highly relevant as most privacy research undertaken to date focuses on consumer related concerns. Very little research exists examining the role of brand reputation and online privacy. Practical implications of this research include gaining knowledge as to how to minimise online privacy concerns; improve brand reputation; and provide insight on how to reduce consumer resistance to the collection of personal information and encourage consumer opt-in.

The architecture of the psychiatric milieu

The last time a peer-reviewed volume on the future of mental health facilities was produced was in 1959, following a symposium organised by the American Psychological Association. The consensus was easy enough to follow and still resonates today: the best spaces to treat psychiatric illness will be in smaller, less restrictive units that offer more privacy and allow greater personalisation of space – possibly a converted hotel (Goshen, 1959). In some way, all those ideals have come to pass. An ideal typology was never established, but even so, units have shrunk from thousands of beds to units that typically house no more than 50 patients. Patients are generally more independent and are free to wander (within a unit) as they please. But the trend toward smaller and freer is reversing. This change is not driven by a desire to find the ideal building nor better models of care, but by growing concerns about budgets, self-harm and psychiatric violence. This issue of the Facilities comes at a time when the healthcare design is increasingly dominated by codes, statutes and guidelines. But the articles herein are a call to stop and think. We are not at the point where guidelines can be helpful, because they do not embody any depth of knowledge nor wisdom. These articles are intended to inject some new research on psychiatric/environmental interactions and also to remind planners and managers that guidelines might not tackle a core misunderstanding: fear-management about patient safety and the safety of society is not the purpose of the psychiatric facility. It is purpose is to create spaces that are suitable for improving the well-being of the mentally ill.

Reputation model based on rating data and application in recommender systems

This thesis introduced two novel reputation models to generate accurate item reputation scores using ratings data and the statistics of the dataset. It also presented an innovative method that incorporates reputation awareness in recommender systems by employing voting system methods to produce more accurate top-N item recommendations. Additionally, this thesis introduced a personalisation method for generating reputation scores based on users' interests, where a single item can have different reputation scores for different users. The personalised reputation scores are then used in the proposed reputation-aware recommender systems to enhance the recommendation quality.