A fuzzy ERM for extracting and modeling uncertain spatial expressions in text

Abstract

The information and knowledge sharing era is exploding with information that people are continuously sharing over various sources across the globe. This has made the retrieval a difficult task. There are plenty of documents everywhere about anything and everything a human mind can think about. All this information is mostly presented transferred and shared using natural language. The biggest challenge and research area has been to enable machines understand and decipher what has been communicated to it through natural language. There are document classification systems that classifies and groups the documents that are speaking about the same concept. But the same type of classification is not successfully handled if it happens to be based on spatial keywords. This is due to the inherent ambiguity and uncertainty that is associated with the spatial terms found in natural language descriptions. Text documents imply the usage of natural language and as such it yields to explicit vague fuzzy descriptions involving linguistic terms such as near to, far from, to the east of, very close and also implicit vague spatial references. Event reporting in case of disasters or in case of special occasions is also generally done using free form text rather than structured methods since it allows more detailed descriptions to be added in. Much of these afore said text documents acting as an information source and the query posed by the user implicitly have a geographic or spatial reference component present in it. This logically leads to the conclusion by the previous studies that more than 80% of the searches are pertaining to geographic locations. Fuzzy logic is an extension to the Boolean crisp logic to accommodate for the fuzziness of an element belonging to a set. This paper studies the feasibility of fuzzy logic techniques in resolving the spatial uncertainty in text and presents a Fuzzy ERM (Extraction, Resolving and Modeling) architecture for the same.

SNEToolkit: Spatial named entities disambiguation toolkit

“…We distinguish three common cases • Case No. 2 3 : a spatially named entity can designate both a place and a nonplace or generally refers to places named after people (e.g., Washington);…”

“…In [2], the authors include a similarity measure of the basics rules of Geonames by using the Levenshtein distance measure to compute the similarity between the extracted place name in the text and the others proposed by the Geonames database. The authors in [3] also applied fuzzy logic techniques to resolve spatial ambiguities. Some approaches, such as [4], use knowledge graphs or gazetteers [5] to disambiguate named entities.…”

“…• resolveByFuzzyMatch 3 : Fuzzy Matching (also referred to as approximate string matching) is a technique that helps identify two elements of text, strings, or entries that are approximately similar [21]. Similar to [3], we introduced these techniques as new strategies with the Geonames API to address the NotAsSNE set to solve problems such as typing problems (e.g., United States instead of United States and incomplete problems (e.g., Angeles instead of Los Angeles.…”