GALE: Global Accessibility to Local Experience

在过去的15年里，已经产生了推荐系统，其目的是根据用户的个人资料，她/他作为客户/网络用户的过去历史，甚至她/他在社交网络中的朋友社区，向个人用户建议在互联网虚拟空间中出现的机会。此外，Facebook等在线社交网络的兴起为推荐系统提供了一个新的信息来源：用户社交网络。互联网接入现在变得越来越移动化，智能手机正在以一种越来越复杂的方式改变人们与地点和彼此互动的方式。智能手机正开始影响用户在旅途中获取信息和接受建议的方式。更具体地说，目前正在开发创新的推荐系统，利用基于gps或其他位置敏感的信息，通过智能手机将其与个人用户联系起来。基于位置的第二代推荐系统带来了一系列完全不同的问题，这些问题不仅与用户的知识（她或他的“个人资料”）有关，还与地点有关。对地方的了解可以通过导游、教科书和旅行报告来获得，也可以通过直接体验来获得。这些方式在本质上是完全不同的。前者可以在全球范围内访问（每个人都可以从远处获得），并且相对较快地获得，特别是在互联网时代。后者只能在当地获得（人们需要在这个地方才能获得），它是由生活在这个地方的人通过个人的当地互动产生的，只有在长期的互动和相互信任的个人关系的构建之后才可以获得。当访问一个新地方时，您必须仅依赖全局信息来导航该地方并访问其资源。相反，如果你是当地人，你对这个地方的了解主要是通过你个人与邻居们每天都在做什么以及他们最喜欢的地方的长期交流建立起来的；因此，你不仅会依赖当地的知识，而且你也会通过在当地的互动，为你的邻居在他们的互动中使用的信息的形成和不断的重新塑造做出贡献。如果我们将当地长期产生的知识称为“邻里知识”，我们可以说，当地人们在某个地方所做的事情在某种程度上取决于他们获得邻里知识的程度。第二代推荐系统允许“全球”地点用户，即访问一个地方的人，他们对这个地方本身没有经验，可以访问“全球”可用的信息。然而，在这些系统中，大量的信息仍然没有被利用，因为地理和社会只是以一种肤浅的方式“相遇”：换句话说，系统没有利用任何关于当地“社区”过去和“此刻”对该地点的特殊使用的信息。由于foursquare或Gowalla等基于位置的社交网络的病毒式传播，社区知识信息现在变得越来越容易获得，现在有可能探索第三代推荐系统，其中关于该地点过去如何被使用（历史使用）或在查询时刻被用户社区使用（实时使用）的知识是系统的关键元素。GALE项目背后的主要动机是开创这样的第三代推荐系统，这将使快速增长的“全球”城市用户有可能访问一个级别的信息，即社区知识，这些信息本来是无法在全球存储库中访问的，并且是实时的。

Recommender Systems have been generated in the past 15 years with the aim to suggest to individual users opportunities arising in the virtual space of the Internet on the basis of the individual profile of the user, her/his past history as a customer/web-user and even her/his friends' community in social networks. Further, the rise of online social networks such as Facebook has allowed for a new source of information to be exploited by recommendation systems: the user social network. Internet access is now becoming increasingly mobile and smart phones are changing the way people interact with places and with each other in an increasingly complex manner. Smart phones are starting to impact the way users access information on the go and receive suggestions. More specifically, innovative recommender systems are currently being developed to exploit GPS-based or other location-sensitive information, associated on-the-go to individual users through smartphones. This second generation of recommender systems, by being location-based, pose an entirely different set of problems which not only have to do with the knowledge of the user (her or his "profile"), but also with that of the places. Knowledge of places can be achieved by means of guides, textbooks and journey reports, or by direct experience. These ways are quite different in nature. The former is globally accessible (everybody can get it from afar) and relatively fast to obtain, especially in the age of the Internet. The latter is only locally accessible (one needs to be in the place to access it) and, being generated by those living in the place through personal local interaction, it becomes accessible only after long-term interactions and the construction of personal relationship of mutual trust. When visiting a new place, you would necessarily rely only on global information to navigate the place and access its resources. Conversely, if you are a local, your knowledge of the place is mostly constructed through your personal long-term exchange with what all your neighbors are doing every day and with their favorite places in the neighborhood; as a result, you not only would rely on local knowledge, but you would also contribute - by interacting locally - to the formation and continuous re-shaping of the information used by your neighbors too in their interaction. If we name the long-term, locally generated knowledge of the place "neighborhood knowledge", we can say that what people locally do in places is in one way or another dependent on the extent to which they have access to the neighborhood knowledge.The second generation of recommender systems allows "global" place-users, i.e. people visiting a place who are not experienced with the place itself, to access "globally" available information. However, a good deal of information is still not exploited in these systems, as the geographic and the social only "meet" in a superficial way: in other words, the system does not take advantage of any information about the particular use of the place that local "communities" have done in the past and do "at the moment". As neighborhood knowledge information is now becoming increasingly available through the viral expansion of location-based social networks such as foursquare or Gowalla, it is now possible to explore a third generation of recommender systems, where knowledge about how the place had been used in the past (historical use) or is used at the moment of the inquiry (real-time use) by communities of users is the key element of the system. The main motivation behind the GALE project is to pioneer such third generation recommender systems which would make it possible for the rapidly growing population of "global" city users to access a level of information, that of the neighborhoods knowledge, which is inherently inaccessible to global repositories, and to do that in real time.