Mikko Terho

Failure prediction based on log files using Random Indexing and Support Vector Machines

Research problem: The impact of failures on software systems can be substantial since the recovery process can require unexpected amounts of time and resources. Accurate failure predictions can help in mitigating the impact of failures. Resources, applications, and services can be scheduled to limit the impact of failures. However, providing accurate predictions sufficiently ahead is challenging. Log files contain messages that represent a change of system state. A sequence or a pattern of messages may be used to predict failures. Contribution: We describe an approach to predict failures based on log files using Random Indexing (RI) and Support Vector Machines (SVMs). Method: RI is applied to represent sequences: each operation is characterized in terms of its context. SVMs associate sequences to a class of failures or non-failures. Weighted SVMs are applied to deal with imbalanced datasets and to improve the true positive rate. We apply our approach to log files collected during approximately three months of work in a large European manufacturing company. Results: According to our results, weighted SVMs sacrifice some specificity to improve sensitivity. Specificity remains higher than 0.80 in four out of six analyzed applications. Conclusions: Overall, our approach is very reliable in predicting both failures and non-failures.

Using RDF Data as Basis for 3D Window Management in Mobile Devices

Abstract Computer users are commonly familiar with the WIMP (Windows, Icons, Menus and Pointing device) paradigm that has been the dominant design for decades. Despite its common use, the WIMP paradigm has a fundamental problem: it clutters the precious screen space with a plethora of open windows. The cluttering becomes an even bigger problem when using mobile devices that have smaller screens than conventional computers. Furthermore, accuracy issues commonly arise with touch screens. In this paper, we are introducing a 3D window manager for mobile devices which aims at solving the above issues by providing simple yet powerful interaction mechanisms and graphics as well as the use of only RDF data instead of application specific content. For improved user experience the user interface has been designed with playful use in mind, and this gives the system unique look and feel.

Scriptable web applications as first-class citizens

Abstract Mashups that combine data from numerous web sites into an integrated experience have rapidly become state-of-the-practice way to develop new compelling applications. However, their use has been somewhat restricted, since it is common that such applications can only be run inside the browser, not in a fashion where they would be directly associated with operating system concepts. Consequently their access to the resources of the hosting system has been limited, which forms a practical obstacle for numerous applications that could benefit from e.g. context information available in mobile context. The limitations are not something that is a fundamental restriction of web technologies, but only related to the existing implementations. In this paper, we show how we have implemented a mashup desktop, where applications that download data – and even complete web applications – from the web can be used in the same way as thumbnails are commonly used. Furthermore, instead of being static, the thumbnails can be alive in a fashion that has been made popular by newer desktop systems, such as Windows 7. The implementation is based on our earlier research on composing applications using JavaScript, the Qt framework, and the K Desktop Environment, whose innovative combination has been an enabler for this work.

Toward a cloud-based mobile device: Scriptable web applications as first-class citizens

The use of applications that are downloaded on the fly has been somewhat restricted. Such applications can often only be run inside the browser, not in a fashion where they would be directly associated with operating system concepts. Consequently, their access to the resources of the hosting system has been limited, which forms a practical obstacle for numerous applications that could benefit from, for example, context information available in mobile setting. Furthermore, due to the design of the run-time systems, web applications are commonly run in the same process, and the protection from other programs within the same process is next to none. These limitations are not something that is a fundamental restriction of web technologies, but only related to the existing implementations. In this study, we show how we have implemented a mobile device, where scriptable applications that download data—and even complete web applications—from the web can be used in the same way as native applications are commonly used. Furthermore, instead of being static, the applications can be alive even when minimized, in a fashion that has been made popular by newer desktop systems, such as Windows 7. The experiences presented in the study are a result of a long-term research project where a new look to the design of a mobile device was taken. Over the project, four milestones can be identified in the work. These are in chronological order: (1) the design of the Lively for Qt system, (2) porting the Lively for Qt system to a desktop of a mobile device, (3) definition of a process model that protects applications from one another when needed, and (4) the introduction of declarative graphics support that liberates the developers from numerous low-level details.

Tool Support for Constructing Mobile Mashups

The ability to instantly publish software worldwide, and the ability to dynamically combine data, code and other content from numerous web sites all over the world has opened up entirely new possibilities for software development. In web terminology, a web site that combines (“mashes up”) content from more than one source into an integrated experience is referred to as a mashup. At present, the development of mashups usually relies on the tools for composing server-side software, and off-the-shelf browser is commonly assumed as the runtime environment. However, when considering client-side mashups that are well-suited for mobile devices due to local processing and associated interactivity, numerous complications exist. One of these problems is available tool support, which is commonly targeted to desktops and browsers. In this paper, we introduce a tool for developing client-side mashup applications. In spirit, the tool is similar to tools available for mainstream mashup development, but all the actual processing is done on the client side using a special purpose runtime environment.