Mobile communications, already in use by three quarters of the planet population and soon to reach each and every person (Williams, 2008), are rapidly evolving to become providers of ubiquitous broadband connectivity (Ramos et al., 2009). Their further evolution will lead to an explosion of mobile content and applications (Feijóo et al., 2009), a fundamental part of the expected mega-trends shaping the internet of the future (Reding, 2008). Many examples could be cited: new entertainment content produced and personalised for the mobile environment, productivity applications for mobile workers, or health and education mobile solutions to increase quality of life. Moreover, some web 2.0 models are already being transferred to the mobile environment since the mobile device is “the mean to harness collective intelligence at the point of inspiration” (Jaokar and Fish, 2006).
Similarly to the wired internet, many of these new mobile web models and applications will require access to data and content in an efficient and user-friendly manner. Search engines, which are already the main gateways for more than half of the users connecting to the internet are already becoming the way to reach appropriate content and applications and to provide additional value to services in mobile platforms, as many recent studies show (Cui and Roto, 2008). The growing mobile penetration world-wide and the increasing mobile broadband availability are additional arguments justifying the increasing demand for effective search tools adapted to the mobile environment.
Mobile search will give added value to users when the results it provides match with their personal expectations, which requires, in turn, a simple-to-use end product or service which is a combination of key technologies such as –search algorithms, displays, context-awareness, wireless sensors and cognitive techniques– and also viable business models.
In the future, web search will face at least two major challenges. One of them is to improve efficiency of retrieving relevant digital content in all formats, audio-visual in particular. While at the early days of the internet, information was predominantly text based, more and more audiovisual multimedia content is now available. In addition, peer-to-peer file sharing networks have been a significant and widely used tool for creating, storing, and exchanging multimedia content on the internet for more than a decade. The second challenge is to retrieve relevant information in a range of platforms, including mobile. These two challenges are not independent. Mobile internet is likely to follow a similar development pattern in content retrieval to the web, namely audio-visual information gradually gaining more relevance to text based one.
Applying this logic, mobile search would result in a mere translation or adaptation of the established PC internet search tools to a mobile environment. Even when adding some “mobility” attributes like the location of the user, this would basically result in extending the same approach (and systems and algorithms) to a new platform with its own specific features and limitations. Most of the available literature on mobile search refers to this “transference” of web search to the mobile domain. For instance, Kamvar and Baluja (2007) argue that a typical search session from a mobile device consists simply of formulating and entering a query, browsing the provided search results and viewing the selected result, and Kolmonen (2008) defines the mobile search engine as “a piece of software designed for a mobile device to provide a service, or a portal, through which the user may submit a query (usually by entering a set of keywords) and get a list of results matching the search criteria”.
However, from a mobile user’s perspective a lot of interesting and relevant data and content can only be found on the internet with the help of contextual information. Such information can be derived from the mobile device, from the surrounding environment and even in the profile and (past, current) behaviour of the user. In addition, it is worthwhile recalling that mobile search is operated from a different device with which we have a much more personal and intimate relationship. Therefore, the mobile device will truly become a tool to bind together the real and virtual worlds (Feijóo et al., 2009a). As a result, mobile search can go further than conventional web search and it should not be just an adaptation of existing internet search solutions to the mobile domain: it should also include new developments that make use of all of these other types of information to improve the meaningfulness and relevance of the search results and/or to provide a different and more valuable user experience.
Along these lines Zoller (2007) has identified ten attributes, from display and input to connection and reach, where web and mobile search significantly differ. They are summarised in Table 1.
From a techno-economic point of view, managing and exploiting the differences between “traditional web search” and “future mobile search” is one of the biggest challenges that mobile search providers face.
| ATTRIBUTE ONLINE | PC MOBILE | PHONE |
| Screen size | Large | Very small |
| Input capabilities | Good | Limited |
| Personalisation and targeting | Reasonable | Very good |
| Connection speed | Fast and improving | Reasonable and improving |
| Site optimisation | Good | Poor but improving |
| Localisation | Reasonable | Very good |
| Consumption patterns | Extended, stationary | Short, on the move |
| Pricing | Flat rate | Metered (but changing) |
| Degree of openness | Completely open | Traditionally closed (but changing) |
| Reach | Significant | Huge |
Mobile search is about much more than the mere translation of present web search to the mobile domain, although most of the available literature has focused thus far on this process. Given the novelty of the domain, there is not yet a commonly accepted terminology in the literature. For instance, Morris (2008) introduces some categories, by making a distinction between the process and product of mobile search, while other observers make a distinction rather on technological grounds. The categorisation we propose follows a market rationale. Classification criteria include attributes like the reach of the search, the input parameters, or the features added to the search. Such attributes cannot be clustered into a single property, thus not offering always clear-cut boundaries between categories, but they help to visualize some market considerations. Table 2 presents a classification scheme together with their technology requirements and their implications per category.
|
MOBILE SEARCH TYPE |
CRITERIA FOR CLASSIFICATION |
REQUIREMENTS |
COMMENTS |
|
On device |
Reach |
Typically software preinstalled in the device making use of non-standardised stored information |
Retrieves information stored in the device |
|
On portal |
Reach |
Typically software preinstalled in the device and bundled with operator/provider services. It is usually provided as a “white label”. |
Typically derived from a “walled garden approach”. In disuse. |
|
Off portal (open) |
Reach |
Typically web search engine with result presentation adapted to mobile environment specific features (incl. using context info or presenting sites optimised for mobile devices). |
Typically available in smartphones and requires a mobile broadband connection |
|
Meta-search |
Method / Input |
Blends results from several search engines to improve results in the mobile environment and to include on-portal content |
It could help to build context-aware search |
|
Social network based |
Method |
Social tagging and folksonomies required |
Highly complementary to any other type |
|
Messaging service based |
Input |
Uses SMS or MMS. Requires a short number to send the request. Could include human response to query. |
Interface with limitations due to length and format |
|
Voice based |
Input |
Typically similar to the above including in addition a speech recognition system |
Nowadays still a slow and inaccurate interface in addition to above limitations if results provided in text |
|
Multimedia (audiovisual and or audio) |
Input |
Uses the camera and or the micro in the mobile phone to query from a picture or sound. Results provided can be standard web pages and/or audio / audiovisual information |
Depends on speech / audiovisual recognition and semantic web developments. Today is still mostly based on annotated content |
|
Context-aware |
Features |
Uses context information (location, time, enviroment data, pictures, user data and profile, etc) to provide refined results of the query |
Still under development |
|
Local search |
Features |
Subset of the above using just location for improved results |
Becoming part of the standard offer of search for the mobile environment |