• No se han encontrado resultados

Collective

Intelligence one or more sources.

Here, “collective” refers to any entity constituted by other entities. Of course this definition also raises the question “what is intelligence?” There are numer-ous definitions of this term which will not be discussed here. In our context the most important facet of intelligence is the ability to solve problems.

When Does Collective Intelligence Occur?

In the context of this thesis, we are interested in Collective Intelligence as a means of supporting the access to digital resources. Thus, we need to identify

40See http://cci.mit.edu

3.6 New Phenomena, Potentials, and Challenges

which prerequisites have to be met when aiming to harness Collective Intelli-gence.

Following our notion of Collective Intelligence, we first of all need “a collec-tive, which arises from many sources.” Such sources in our context are pieces of information, derived from actions and contributions by individuals (e.g., tags annotated by users in a social bookmarking system). Collecting and processing such information was usually a complex and time-consuming task. Nowadays, by using online tools and systems, it is no longer a technical problem to gather and process data from and about millions of users [Seg07].

Yet, just having the data at one’s disposal is not enough when aiming to harness Collective Intelligence. Of course not every contribution that is pro-duced from any collective provides information from which intelligence arises.

In [Sur04], Surowiecki identifies four conditions that characterise what he calls a

“wise crowd”, i.e., a number of individuals that contributes information allow-ing the harnessallow-ing of Collective Intelligence:

1. “Diversity of opinion – each person should have some private information, even if it is just an eccentric interpretation of the known facts.” When such a di-versity does not exist, contributions of individuals will not be different enough to allow that intelligence arises out of them (see [LDP08] for a de-tailed discussion on diversity in open social networks).

2. “Independence – people’s opinions are not determined by the opinions of those around them.” This condition is important, because otherwise people might be influenced too much by other’s actions (social proof is a well known psychological phenomenon that occurs when the behaviour of others is considered as a model for the own behaviour). A lack of independence can lead to the unwanted occurrence of what is often referred to as cascade effects. Especially in the context of web applications, the use of aggregate displays (i.e., displaying information in a way that shows what other users did) influences people’s opinions [Por07].

3. “Decentralisation – people are able to specialise and draw on local knowledge.”

To ensure that the diversity of people’s contributions also reflects all infor-mation that is required for a result that is considered as intelligent, decen-tralisation is an important aspect. The infrastructure of the World Wide Web provides a good basis to ensure this, as it is accessible from basically everywhere.

4. “Aggregation – some mechanism exists for turning private judgements into a collective decision.” Depending on the concrete scenario, the contributions

by the crowd will have to be aggregated in an according way. Well-known examples in the area of Web 2.0 are to rank resources by ratings or number of views, and to aggregate the tags people used in a tag cloud.

Potentials and Challenges

Fulfilling the above mentioned criteria to allow the harvesting of Collective

In-Challenges

telligence is a non-trivial issue. It requires mechanisms that:

• attract and motivate users, and that allow decentralised contributions in a way that fosters independence and diversity of these contributions,

• aggregate and process the information contributed by the users in way that produces an added value.

Once this is realised, Collective Intelligence can be a very important source

Potentials

for supporting the access to digital resources. This concerns two aspects:

1. The provision of social resources (e.g., Wikipedia articles) and social meta-data (e.g., tags and ratings) that might be retrieved as relevant output of an information retrieval or information seeking process.

2. The creation of social metadata that allows to harvest Collective Intelli-gence for

• filtering, searching and ranking, and for

• recommendations (e.g., using collaborative and social filtering).

3.6.4 Crowdsourcing

As already stated, the World Wide Web offers an infrastructure and technol-ogy allowing to reach very large audiences and to aggregate contributions of users on a large scale. In the last years, these capabilities were used by more and more organisations and platforms to outsource tasks formerly conducted by themselves to potentially any user – so called Crowdsourcing. The term Crowd-sourcing was first coined by Jeff Howe in the June 2006 issue of Wired magazine [How06b]. He provides the following definition:

“Crowdsourcing is the act of a company or institution taking a function once performed by employees and outsourcing it to an undefined (and gen-erally large) network of people in the form of an open call.” [How06a]

3.6 New Phenomena, Potentials, and Challenges

Howe further details that Crowdsourcing “can take the form of peer-production (when the job is performed collaboratively), but is also often un-dertaken by sole individuals’ and that “the crucial prerequisite is the use of the open call format and the large network of potential laborers.” This also makes clear that Crowdsourcing may not be confused with Collective Intelligence – the contributions of people in a Crowdsourcing process might results in the occur-rence of Collective Intelligence, but this is not required.

To illustrate the Crowdsourcing concept, we will now provide some exam-ples:

Goldcorp Challenge: In 2000, the Canadian gold mining company Goldcorp initiated the “Goldcorp Challenge”. Participants were encouraged to ex-amine parts of Goldcorp’s geologic data, and to submit proposals identify-ing potential targets where the next 6 million ounces of gold will be found.

As prize money, Goldcorp offered more than 500 thousand US Dollars to the 25 finalists that identified the most gold deposits. Goldcorp managed to attract “more than 400 online prospectors from 51 countries registered as Challenge participants” [Gol01, p.2]. The submitted solutions identified a total of 110 deposits, confirming many of Goldcorp’s suspected deposits, but also identifying several new ones [Bra08].

SETI@home: SETI@home41 is hosted by the Space Sciences Laboratory at the University of California, Berkeley. It uses Internet-connected computers to search for extraterrestrial intelligence. People can participate by down-loading a program that automatically downloads and analyses radio tele-scope data.

Mechanical Turk: Amazon’s Mechanical Turk42offers to businesses and devel-opers the opportunity to ask workers to complete so-called “HITS” (Hu-man Intelligence Tasks). Workers on the other hand can choose according HITS they want to work on and get paid for. As of November 2008, Me-chanical Turk offered more than 110 thousand of such HITS.

Adobe knowhow: Adobe knowhow43 is a technology preview that uses Deli-cious as a platform to collect user generated content that is then provided in the Adobe Creative Suite 3. Users can suggest new content like tool descriptions or tutorials.

41See http://setiathome.berkeley.edu

42See https://www.mturk.com/mturk/welcome

43See http://labs.adobe.com/technologies/knowhow

Potentials and Challenges

Crowdsourcing can be applied for any kind of task that could otherwise only be

Potentials

realised with a lot of work by specialised experts or employees. In our context, we have to consider the social resources and social metadata created explicitly with the interaction possibilities as introduced in Section 3.6.1. As a result, we can derive the following tasks to apply Crowdsourcing:

Provision of digital resources: The creation and the provision of digital re-sources that meet some specified needs (e.g., courses in an eLearning con-text, manuals that provide information about how to use a product, or contents related to a certain topic) is often a time-consuming process that requires a lot of expertise. As mentioned in the Adobe knowhow example, such tasks can sometimes also be carried out by end users.

Creation of metadata: When the amount of resources that have to be anno-tated with metadata is very large (and eventually grows each day), tradi-tional approaches to describe the content usually cannot be applied. This especially holds for the World Wide Web, and with the rise of Web 2.0, even more resources are being created. As Guenther and McCallum state: “The rapid proliferation of digital resources demands both rapidly produced descriptive data and the encoding of more types of metadata” [GM03, p.12]. In traditional environments such as libraries, the metadata was usu-ally created by experts such as librarians or archivists. Such metadata pro-vides information such as the title and author of a resource, or a classifica-tion. But not only the Crowdsourcing of this kind of metadata can be used to allow for an efficient retrieval of resources. As already stated, also social metadata such as ratings or tags can be a very useful source.

In order to realise a successful approach based on Crowdsourcing, one faces

Challenges

two main challenges:

• A sufficient amount of users with appropriate resources to contribute to the solution of the given task or problem has to be addressed. Although we can reach potentially any user by using the infrastructure of the World Wide Web, this obviously is not enough – the attention of according tar-get groups has to be attracted, and the users have to be motivated to con-tribute. This might be realised by offering rewards such as money, but also with game-based approaches as introduced by Luis von Ahn (for more in-formation see [vAD04, vAGK+06]).

• We must of course offer the opportunities that allow to contribute infor-mation in the desired way.

3.7 Summary and Conclusion

3.7 Summary and Conclusion

The opportunities for accessing, creating and sharing information have changed a lot with the rise of the World Wide Web, and especially with social media tools available in what we call the Web 2.0. Users turned more and more from con-sumers to producers. Without requiring any special expertise or permissions, people can actively contribute and share information as well as communicate with others on a world-wide scale. This paradigm shift from traditional to social media entails several phenomena that provide potentials as well as challenges when aiming to support the access to digital resources.

In this chapter, a holistic framework with a clear terminology and structuring of entities in Web 2.0 applications was provided, in order to then identify these potentials and challenges.

Table 3.5 provides a summary of the identified potentials and challenges, with a special focus on the creation and provision of social metadata and social re-sources. If direct links exists to the characteristics of an ideal setting as presented in Section 2.3.1, the respective characteristics are mentioned accordingly.

Phenomenon Potentials Challenges

3.7 Summary and Conclusion by the users in way that produces an added value

• Attracting the attention of a sufficient amount of users

Table 3.5:Web 2.0 related potentials and challenges to support access to digital resources

C HAPTER 4