• No se han encontrado resultados

HAY QUE HABLAR CLARO

In document EL AMIGO DEL PUEBLO (página 80-83)

In this chapter, we observed search performance and behaviour across three different sizes of screen. Our findings indicate that search behaviours and usability vary across screens, although participants exhibited a similar search time and accuracy. Glancing through the results reported in this chapter, one might conclude that there is no need to change anything in SERPs because there is no significant difference for search time and accuracy. However, if we could create better design for SERPs for each type of screen by understanding search behaviour, this could provide users with a better search experience.

We know that adding information into snippets clearly improves the search per- formance for informational tasks on desktop monitors [Cutrell and Guan, 2007]. However, we need to consider the effects of screen size on mobile devices. We sug- gest several possible ideas for designing the interface for web searches on the screens for each mobile device, bearing the limitations of our study in our mind. First, par- ticipants in our experiment did not read as many links as there were visible on the screen, less scrolling on the large screen. Therefore, for the screen on a phablet that is 5.5 inches or similar, one promising idea is todisplay a knowledge graphthat displays information regarding the keywords that may be helpful for search performance, as shown in a previous study [Lagun et al., 2014]. Because the screen has enough space (for about six links) to display the KG with several of top links on the initial SERP, one possible design would be to locate the KG at the top of the screen instead of the top three links, moving them below the KG to reduce the effect of decreasing search speed that occurs when the KG is not relevant, as shown in Lagun et al. [2014].

Second, our users reported a low usability for the small screen and had a slow reading performance for each link. The best way to improve satisfaction, and reduce scrolling, would be to widen the screen. This is clearly impractical. Therefore, we may suggest three ideas for earlier-type smart phones (3.5 inches), which could con- tribute to an increase in satisfaction as well as search speed: making the best use of peripheral vision and reducing font size to display more contents, and embedding page up and down button on the interface or horizontal page changes instead of a vertical scroll function, as suggested by Jones et al. [1999]. Additionally, displaying only one link that has rich information in a snippet along with one of the above functions on the smallest screen is a promising design to reduce the time consumed by scrolling.

Third, on the medium-sized screen, the subjects exhibited a faster reading speed with hesitant eye-movement to make a decision among the top links. Consequently,

§4.4 Conclusions and future work 47

we expect thatenriching the content of top links by showing longer snippets [Cutrell and Guan, 2007] could reduce the hesitation on the recent smart phones (4.7 inches). That is, we could display only three links with rich snippets instead of showing four links. This could reduce the hesitation behaviour by providing additional information to help the user choose a relevant link. We focused on this in study three, see Chapter 6. In addition, for general mobile devices, we recommend providing a small mark indicating that an item on a SERP links to a mobile-optimised page instead of a full-size page for desktops as mentioned in our previous work [Kim et al., 2015], and as initially suggested by Jones et al. [2003]. The Google mobile search engine provides the indicator by marking with “Mobile-friendly” in front of each snippet on SERPs, if the links connect to mobile-optimised pages. This may reduce the user time cost on SERP as well as on web documents after selecting a link, and improve search all sizes of small screens.

We believe that the above recommendations from this study may contribute to better SERP presentation designs for each mobile screen size.

Chapter5

Study Two: Pagination versus

scrolling

One possible idea for the SERP design for small screens in the previous chapter 4 (Study One) indicates that ‘embedding page up and down button on the interface or horizontal page changes instead of a vertical scroll function’ would provide better user experience in mobile web search. Horizontal pagination is used as one viewport control type on several mobile applications. However, mobile search engines still provide only vertical scrolling to show the contents beyond the page fold. In this chapter, we address the second study about the effect of horizontal pagination as one of the future work from the study one.

5.1

Introduction

Although several studies have indicated that the use of a scroll function is closely related to search performance and behaviour (e.g., increasing search time and some special interests on links around a page fold) [Cutrell and Guan, 2007; Granka et al., 2004; Joachims et al., 2005; Kelly and Azzopardi, 2015], people must still scroll through SERPs to see beyond the page folds.

Compared to scrolling on desktop monitors, screens for smartphones need to be scrolled more because they have less visible space, and the effect of scrolling may be more important because users need to cover part of the screen with their fingers while scrolling.

To initiate a scroll event on a touch screen, users drag their fingers vertically to produce a similar effect to spinning a mouse wheel, holding and dragging the scroll bar, or using the page up/down keys on desktops. To the best of our knowledge, vertical scrolling is the only option provided by all search engines to control SERPs on touch-enabled mobile devices. However, people are familiar with turning pages

50 Study Two: Pagination versus scrolling

Figure 5.1: Example of two subsequent pages of a SERP with the horizontal control type.

horizontally when reading a book, either on paper or via an eReader, with discarding unwanted applications or web pages on smartphones [Warr and Chi, 2013], with the start screens for smartphones (pages of icons or tiles), and with controlling appli- cations which provide a switching interface using horizontal swiping gestures (e.g., weather, stock price and shopping). Therefore, horizontal swiping for mobile web search could also have been considered by interface designers, and we wondered what would happen, in terms of search performance and behaviour, if users could swipe horizontally (paginate), as shown in Figure 5.1. Although most SERPs have buttons for moving to the next page, this function is somewhat different from a hor- izontal swiping interface. With new HTML and CSS standards, it is now relatively simple to load multiple SERPs, and the pre-loaded pages can be hidden before being called up. This means that there is no additional loading time to display the same

In document EL AMIGO DEL PUEBLO (página 80-83)