Layout as a Service (LaaS): A Service Platform for Self-Optimizing Web Layouts

In Proceedings of the 20th International Conference on Web Engineering, ICWE 2020.
User Interfaces group, Aalto University
LaaS teaser
A new approach to self-optimizing web pages personalize pages and improve usability automatically
  • Improves usability: Central aspects of the web layout are reoptimized for a user, making it easier to find and click important items. In previous studies, improvements to usability have been between 5% and 25%.
  • Lightweight: Assumes access to click data only, easy to customize by changing design objectives only. No need to manually define templates, the whole grid design can be decided by the optimizer just like a human designer would. Installation is simple, requires a single line to be added by the service owner.
  • Controllable: A control panel to manage the service and its settings is offered.
  • Based on recent results: The method is based on combinatorial optimization and a recent breakthrough in applying it in graphical layout design.
Abstract

To personalize a web page, case-specific rules or templates must be specified that define the visuospatial layout of elements as well as device-specific adaptation rules for an individual. This approach scales poorly. We present LaaS, a service platform for self-optimizing web layouts to improve their usability at individual, group, and population levels. No hand-coded rules or templates are needed, as LaaS uses combinatorial optimization to generate web layouts for stated design objectives. This allows personalization to be controlled via intuitive objectives that affect the full web layout. We present an extensible architecture and solutions for (1) layout generation using integer programming, (2) data abstractions to mediate between browsers and layout generators, and (3) page restructuring. Moreover, we show how LaaS can be easily deployed as part of existing web pages. Results demonstrate that our approach can produce usable personalized web layouts in diverse scenarios.

Materials

Materials from ICWE 2020 (Helsinki, Finland online).

Publication
LaaS paper

PDF, 2.4 MB
Markku Laine, Ai Nakajima, Niraj Dayama, Antti Oulasvirta. 2020. Layout as a Service (LaaS): A Service Platform for Self-Optimizing Web Layouts. In Proceedings of the 20th International Conference on Web Engineering (ICWE '20). 

                            
@inproceedings{laine2020_laas,
    author={Laine, Markku and Nakajima, Ai and Dayama, Niraj and Oulasvirta, Antti},
    editor={Bielikova, Maria and Mikkonen, Tommi and Pautasso, Cesare},
    title={Layout as a Service (LaaS): A Service Platform for Self-Optimizing Web Layouts},
    booktitle={Web Engineering},
    year={2020},
    publisher={Springer International Publishing},
    address={Cham},
    pages={19--26},
    isbn={978-3-030-50578-3},
    doi={10.1007/978-3-030-50578-3\_2}
}
Contact

For questions and further information, please contact:

Markku Laine

Email:
markku.laine (at) aalto.fi

Acknowledgements: This work was funded by Technology Industries of Finland project SOWP, EU Project HumaneAI, and the European Research Council (ERC; grant agreement 637991).