Learning boundaries of vague places from noisy annotations

Abstract

What ordinary people mean by places may differ dramatically from what experts consider them to be. This is especially evident in how people talk about places in social media, where 'Los Angeles', for instance, could include areas well outside of the city or even in another county. In order to make best use of the information in social media, we need to understand what people mean when they refer to a place. Social annotations provide valuable evidence for harvesting knowledge about places, e.g., learning their boundaries and relations to other places. However, social annotations are noisy, and this can dramatically distort the learned boundaries. In this paper we propose a method that exploits the distinctive property of social annotations -that it is created by many people -to filter out noise. Using a large data set extracted from Flickr we show that our crowd-based noise filtering method can learn accurate boundaries of places, including vague places.

Extracting and understanding urban areas of interest using geotagged photos

“…A number of studies have been conducted based on Flickr data. For example, Hollenstein and Purves (2010), Intagorn and Lerman (2011), as well as Li and Goodchild (2012) have used Flickr data to investigate the boundaries of vague place concepts. Keßler et al (2009) and Gao et al (2014) employed Flickr data to enrich gazetteers with new place entries.…”

Colloquial region discovery for retail products: discovery and application

“…Previous studies on similar theme [4,5] have exclusively dealt with extraction of colloquial boundaries for geographic concepts (such as "germany") for which the ground truth is easily available. The region extracted for tag "germany" should overlap with the international boundaries of the eponymous country.…”

“…Previous studies have targeted several different data sources such as photographs, blogs and tweets. [1][2][3][4][5]. The state of the art is established in [5] where the authors first generate a frequency histogram and use an image segmentation-based technique to isolate the characterizing region.…”

“…Uncovering colloquial regions corresponding to entities from geo-tagged data has recently drawn attention of the research community [1][2][3][4]. The current state of the art is established in [5] where the authors develop a technique (SSR) to identify regions for geo-entities.…”

A probabilistic approach to mining geospatial knowledge from social annotations

“…Such knowledge will help us more effectively utilize the massive quantities of user-generated content across a wide spectrum of applications, from responding to disasters, to monitoring the environment, managing resources, and interacting with the world and one another, Existing works have explored various methods for harvesting geospatial knowledge from geo-tagged user-generated content. They use clustering or information theoretic methods to analyze the spatial distribution of tags to extract place semantics [11], place boundaries [5], or suggest locations for photos based on their tags [3,12]. However, extracting knowledge from annotations created by many different people creates challenges that existing works did not address.…”