Named Entity Disambiguation for Maritime-related Data Retrieved from Heterogenous Sources

Abstract

The article concerns integration and disambiguation of data related to the maritime domain. A developed system is described, which collects and merges data about several maritime-related entities (vessels, vessel types, ports, companies etc.) retrieved from different internet sources and feeds the data into a single database. This process is however not trivial. There are few challenges, which need to be faced to successfully conduct it. Firstly, in different sources, entities may be referenced to in different ways, for example, by using different text strings. Additionally, some of these references may be ambiguous, i.e. potentially the reference may point to more than one entity. To enable efficient analysis of data coming from different sources, such ambiguities must be resolved automatically as a preprocessing step, before the data is uploaded to the database and utilized in further computations. The aim of the disambiguation process is to assign artificial, unique identifiers to each entity and then, if possible, automatically assign these identifiers to each data item related to a given entity. In the article, developed methods for resolving such ambiguities are discussed and their evaluation is presented.

Big Data for Anomaly Detection in Maritime Surveillance: Spatial AIS Data Analysis for Tankers

“…Therefore, appropriate methods were developed to alleviate these quality issues. The process of internet sources selection and data fusion was described in detail in separate papers ( [35] and [19] respectively). As a result a vast amount of ancillary data for all types of vessels was acquired, such as tonnage, dimensions, detailed type, built year, builder, home port, data about detentions and inspections of ships as well as data about classification statuses of ships and their affiliation to a classification society.…”

Data analysis in the maritime domain

“…The SIMMO system retrieves data from many sources, and each of these sources may a have different structure and may publish data in a different way. Therefore, a separate Data Acquisition Module (DAM) was developed for each data source (Małyszko et al, 2016). DAMs connect to the data source in a defined manner, send appropriate requests, collect the documents returned, and extract required data.…”

“…Such methods were developed within the SIMMO project in order to address this challenge. They are presented in detail in another paper (Małyszko et al, 2016).…”