• No se han encontrado resultados

y justificación de su empleo.

III.4. Proceso Analítico en Red (ANP)

Usability studies of mobile adaptation conducted by Roto and Kaikkonen (2003) suggests that adaptation engines should preserve, as much as possible, the original structure of the full site version in order to allow users to associate and relate the mobile version with the original one.

We have endeavoured to ensure consistency is retained between devices with our approach. We defined an adaptive page - a structure for a web page (using divs) that allows the same page to be prioritised and delivered to different devices based on users’ preferences (see Chapter 5). We developed a prioritisation engine that prioritises the adaptive page (or the base page) based on the preferences. This approach requires little effort from the developers and users. The web developers need to have their web pages structured using divs; and ensure each item (div) to be prioritised has a unique id within a page. Developers should provide a default items (preferences) for each type of devices. The users could accept the defaults items or set their preferences of items of their interest based on their devices.

Our results show the feasibility of an adaptive page approach to produce similar pages for different devices. Our Prioritisation engine enables users to specify which items they would like delivered to different devices. This is done in such a way as to preserve the page’s parent- child relationships, ensuring the overall layout and structure of the original page is maintained. Our trials’ results showed that this similarity and familiarity helped users to have a good browsing experience, in which they were able to relate and recall their experience with

138 the familiar full site version, thus performing tasks easily. In comparison, users had difficulty with navigating the differently structured pages for Facebook desktop and mobile version. Two other adaptation approaches, PageTailor (Bila et al., 2007) and Proteus (Caetano et al., 2007), are similar to ours and were discussed in Chapter 2. PageTailor is a client-side approach that allows users to show, hide, and re-arrange blocks of content once and is automatically applicable for other pages of similar structure. This only works on mobile devices with Minimo browser. Proteus is a proxy-side adaptation architecture which converts pages into conventional HTML of summarised text with an appended ‘more’ link to expand the page or converts pages into thumbnail depending on users’ preference. Table 9.1 compares the Prioritisation engine to these other approaches.

Table 9.1 Comparison of the Prioritisation engine to other similar approach Prioritisation PageTailor (Bila et al.,

2007)

Proteus (Caetano et al., 2007)

Approach (where)

Server Client / Proxy Proxy

How Show, remove, and re- arrange items of interest, and adds ‘more...’ link.

Show, hide, and re- arrange items of interest.

Convert and compress page to a summarised text and appends ‘more’ link, and/or convert page to a thumbnail.

No re-arranging of item Description Uses the DOM and

XSLT to prioritise the page

Device and browser independence.

Use the DOM and XPath to apply the customisation Device dependent - require devices with Minimo browser.

Uses the DOM to analyse page, and a similar algorithm to the Vision-based Page Segmentation (VIPS) algorithm to highlight content. User preferences

Store the id and ranks of items based on users and devices

Store XPath of content Store profile e.g. compression rate, area of interest

Developer Need to specify prioritisable divs by assigning unique id to each.

Do not need to do anything; and cannot determine (do not have full control of what to customise.

Do not need to do anything; do not have control over what to customise.

User Provided with default page; only need a small customisation if needed

Need to perform the customisation on mobile device

Need to do the

customisation – specify values for items (e.g. font size)

139 Our implementation was based on the assumption that web developers (or other stakeholders involve in decision making in the website development) would easily detect and determine the areas of content to be prioritised (the divs that could be prioritised) and would follow the general recommendation that each id within a page should be unique.

Identifying items based on its ids is a straightforward and an ideal approach. However, while it is based on the HTML recommendation that items should have a unique id, this may not be strictly followed in reality.

While this approach has positive implication to the users, it may introduce extra work for web developers or content providers. Users can specify page’ items and order of appearance based on their preferences. This will ensure that users get only the specified content. On the other hand, content providers need to carefully ensure that items to be prioritised (prioritisable items) have unique ids. In practice, this may add more effort to them.

Alternatively, another approach as used by Bila et al. (2007) and Kao et al. (2009) detects and records the path to the unique items. This approach could overcome the possibility of items not having ids or unique id. This approach uses and stores the XPath expression to identify the items. While this may lessen developers’ effort in assigning unique id, it may not be ideal in all cases in reality as using XPath is only suitable and works well for pages with a structure that is rarely changed.