Geographical Information Retrieval
It's not every day, one queries for "Sports Updates" sitting in a hostel room in Delhi, India and ends up with "How New York Giants performed last night?". What does it return instead, India are 28/0 in a game of Cricket against Bangladesh. Also, a weather inquiry for Hyderabad doesn't return details for Hyderabad, Pakistan.
What leads to such kind of results?
Geographical Information Retrieval or GIR is an addition to general Information Retrieval with emphasis on queries concerning Geographical aspects of the query and in some instances from where the query was made from.
With data access across Mobile devices estimated to contribute about 80% to total data access done on the planet by 2020, GIR is playing a significant role in defining how we access data based on location query is related to. Evidence of such change is apparently evident, with introduction and development of Personal Assistants such as Google Assistant, Siri or Alexa. But again, it is these Mobile devices which have primarily contributed to the addition of Geographical aspects to knowledge bank, also social network sites such as Facebook, Twitter has become portal for such information at large.
How GIR is fundamentally implemented?
Results of GIR is a combined result of similarity results of two different query, one without Geographical details and other with documents with geographical information. The final result is a weighted combination of the two results, with the weight depending upon how strict is the query with location information in need. For example, Query "Restaurants around me" is much stricter to location-specific information than "Recent Developments," where the user may prefer localized news before international news but also wants to be informed about significant developments around the world.
But just returning results is one thing, and returning relevant results is an entirely different aspect. With GIR this raises some severe challenges to researcher and developers.
With data access across Mobile devices estimated to contribute about 80% to total data access done on the planet by 2020, GIR is playing a significant role in defining how we access data based on location query is related to. Evidence of such change is apparently evident, with introduction and development of Personal Assistants such as Google Assistant, Siri or Alexa. But again, it is these Mobile devices which have primarily contributed to the addition of Geographical aspects to knowledge bank, also social network sites such as Facebook, Twitter has become portal for such information at large.
Geographical Aspects of Query "Football" [Source: Frakenplace]
Geographical Aspects of Query "Cricket" [Source: Frakenplace]
How GIR is fundamentally implemented?
Results of GIR is a combined result of similarity results of two different query, one without Geographical details and other with documents with geographical information. The final result is a weighted combination of the two results, with the weight depending upon how strict is the query with location information in need. For example, Query "Restaurants around me" is much stricter to location-specific information than "Recent Developments," where the user may prefer localized news before international news but also wants to be informed about significant developments around the world.
IR: Information Retrieval
GIS: Geographical Information Systems
GIR: Geographical Information Retrieval
But just returning results is one thing, and returning relevant results is an entirely different aspect. With GIR this raises some severe challenges to researcher and developers.
- Extracting Geographical Content within Text documents as well as User's Query.
- Differentiating between places with similar names, for example, Hyderabad, India and Hyderabad, Pakistan.
- Indexing by both Geographical context as well as theme.
- Defining methods to determine weights, i.e., when Geographical content needed to be given more weight and when not to.
Though these challenges can be seen as separate research problems. But, as they are not limited to some thousand documents or cater to only a small fraction of users, these are the problems which impact a large section of society and developments in these fields are revolutionizing how we access data more efficiently. The localization of data concerning location will soon become a necessary thing, and GIR will become a fundamental aspect of IR, not an augmentation.
References :
- Frankenplace
- Christopher B. Jones, R. Purves, A. Ruas, M. Sanderson, M. Sester, M. van Kreveld, and R. Weibel. 2002. Spatial information retrieval and geographical ontologies an overview of the SPIRIT project.
- Jones C.B., Alani H., Tudhope D. (2001) Geographical Information Retrieval with Ontologies of Place.
Comments
Post a Comment