reader type was not found in the dataset. A possible explanation could be that this behaviour was relatively less represented in our dataset, and hence news readers are grouped into two behaviours regarding their habits. Those who are compul- sive/obsessive about accessing news (Trackers - Cluster 1) and those who are more traditional in-depth readers (Reviewers - Cluster 0 and 2). This claim could also be supported by the need for consuming news, particularly created from social media, that was absent in previous years. In the final step of the interpretation, we ex- amined the distribution of the personality traits and NFC across the newly formed clusters (Figure 3.8 and Table 3.12). As can be observed participants who grouped in Cluster 0 (i.e. Tracker/Reviewer) scored higher across the five personality traits except for Extroversion. In relation to the other two clusters, BFI scores were lower, suggesting intermediate levels of each trait. The higher NFC score was observed in Cluster 2 (i.e., Reviewer/Tracker), a finding that is aligned with the definition of a Reviewer type (i.e., a person who seeks for an in-depth understand of news content) and indicates a tendency towards undertaking challenging activities.
3.7
Discussion
This Chapter examined people’s mobile news consumption patterns and defines a News Reader Typology that reflects the different ways people consume and access news content. While previous studies have independently examined different fac- tors that are associated with people’s preferences and choices of news content, this study investigates news reading factors that are related with people’s navigational, reading and contextual behaviour in relation to consumption habits.
An online questionnaire was designed and deployed with the aim of investi- gating people’s mobile news reading behaviour. A descriptive analysis on people’s responses revealed a news reading behaviour that is characterised by more time spent on browsing and skimming stories, one-time reading and less in-depth read- ing. Similar results were reported in a study conducted by Liu (2005) that attempted to investigate reading behaviour in the digital environment. The statistical analysis
performed on the online questionnaire suggested a mobile news readers typology that is characterised by the factors of frequency, total daily reading time, time of the day, location, image captions preference. The factors of browsing strategy and reading style did not show statistical significance but further investigation is needed. Chapter 5 examines both factors with behaviourial data from users’ interactions with a news app.
The News Reader Typology proposed in this Chapter contributes towards the ultimate goal of this research. Firstly, it establishes three news reader personas that are well defined and distinct, which is fundamental for applying personalisation in news apps user interfaces. The three news reader personas are (a) ‘Trackers’, people who follow the news throughout the day and skimming over them; (b) ‘Re- viewers’, people who are daily routine news readers, mainly the traditional daily catch up of the news that sometimes tend to engage with an in-depth understanding of the news content; and (c) ‘Dippers’, people who are liberate surfer news readers, mainly described as casual readers who are looking for specific facts of information without reading everything through keyword-spotting. Secondly, having extracted factors that are statistically significant with news reading behaviour it can inform the design of a mechanism that collects such data from a dedicated news app, i.e. a news app that systematically monitors the seven significant factors from inter- acting with the news app’s user interface. Additionally, aside from the Chapter’s primary aim, an investigation of domain-independent factors (i.e. personality traits and need-for-cognition) in relation to the domain-specific news reading factors was carried out. The analysis showed the added value of personality traits and NFC in identifying a person’s news reader type. Specifically, a Silhouette analysis showed an improvement of 76.64% increase when adding those domain-independent fac- tors into domain-specific factors about news reading behaviour. Furthermore, we showed an ad-hoc interpretation of the newly formed clusters by relating them to the news reader typology. Finally, by analysing those factors, it would be possi- ble to build a user model that is capable of detecting and recognising the user’s stereotypical behaviour of news consumption.
Chapter 4
An Adaptive News Research
Platform
The previous chapter investigated people’s differences in the way they read and con- sume news on their mobile devices through an online questionnaire, which resulted in a News Reader Typology. The News Reader Typology defines three prototypical types of news readers characterised by six discriminating factors, mainly describing how people browse news headlines and how they read news stories.
This Chapter presents the design and implementation of an adaptive news framework that defines a research platform, which seeks to address questions on automatic detection of mobile news reading behaviour and adaptation of the user interface and interaction. The implementation of an adaptive news framework is essential towards the ultimate goal of this thesis, as it will provide the research plat- form to examine the research questions defined in the Introduction. Its ultimate goal is to facilitate the idea of extending beyond what news content people read and access on mobile news apps to how they read and interact with it.
The Chapter presents the architecture of the research framework and discusses its main components from a technical perspective. The framework was designed in a three-tier architecture. The well-known three-tier architecture was favoured in our research framework, mainly because it gives the ability to update the underly- ing technology of one tier without impacting other areas of the application stack, it provides an ease maintenance of the code base, as it keeps the presentation code
separate from the business logic and the data, and it can easily scale up the appli- cation. The framework introduces all the components that in subsequent Chapters will tackle the research questions around adaptivity in news apps. Particularly, it discusses the components in relation to how the automatic detection of a user’s news reader type can be achieved and the automatic adaptation of the user interface and interaction.
The Chapter begins with a motivation and explains the need for a research framework. It then presents the framework’s architecture and highlights its main components through architecture diagrams and technical implementation details.
The architecture presented in this Chapter was presented in the Doctoral Con- sortium of CHI’15 and appeared to CHI ’15 Extended Abstracts. (Constantinides, M. Apps with habits: Adaptive interfaces for news apps. In Proceedings of the 33rd Annual ACM Conference Extended Abstracts on Human Factors in Comput- ing Systems (pp. 191-194). ACM)
Our prototype mobile news app, Habito News, presented in this Chapter has appeared to MobileHCI ’15 as a demo. (Constantinides, M., Dowell, J., Johnson, D., Malacria, S. A research tool to investigate mobile news reading. In Proc. of MobileHCI Adjunct).
4.1
Motivation
The work presented in this Chapter introduces a framework as the research plat- form to investigate how personalisation on mobile news app can be achieved, and extend beyond, existing literature on news content recommendation to incorporate personalisation of the user interface and interaction with news services. It presents the framework that its main components seek to address; questions on automatic detection and recognition of mobile news reading behaviour and questions related to automatic adaptation of the user interface and the interaction.