Assisting Forensic Identification through Unsupervised Information Extraction of Free Text Autopsy Reports: The Disappearances Cases during the Brazilian Military Dictatorship

Martín-Rodilla, Patricia; Hattori, Marcia Lika; González-Pérez, César

doi:10.3390/info10070231

Cited by 4 publications

(3 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In 2019, Martin-Rodilla et al [28] describe the analysis of 3000 textual reports in São Paulo during the Brazilian dictatorship by using unsupervised and supervised approaches, supported by the Linguakit Suite, the Stanford CoreNLP (https://stanfordnlp.github.io/ CoreNLP/ (Accessed on 1 June 2021)) tool, SIEMÊS [29] algorithms, by identifying NEs and relevant terms. This approach tries to set people's information and automate the study of correlations between actors.…”

Section: Literature Reviewmentioning

confidence: 99%

A Graph Database Representation of Portuguese Criminal-Related Documents

2021

View full text Add to dashboard Cite

Organizations have been challenged by the need to process an increasing amount of data, both structured and unstructured, retrieved from heterogeneous sources. Criminal investigation police are among these organizations, as they have to manually process a vast number of criminal reports, news articles related to crimes, occurrence and evidence reports, and other unstructured documents. Automatic extraction and representation of data and knowledge in such documents is an essential task to reduce the manual analysis burden and to automate the discovering of names and entities relationships that may exist in a case. This paper presents SEMCrime, a framework used to extract and classify named-entities and relations in Portuguese criminal reports and documents, and represent the data retrieved into a graph database. A 5WH1 (Who, What, Why, Where, When, and How) information extraction method was applied, and a graph database representation was used to store and visualize the relations extracted from the documents. Promising results were obtained with a prototype developed to evaluate the framework, namely a name-entity recognition with an F-Measure of 0.73, and a 5W1H information extraction performance with an F-Measure of 0.65.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

A Graph Database Representation of Portuguese Criminal-Related Documents

2021

View full text Add to dashboard Cite

show abstract

“…As we said before, the assessors should be familiar with the topic of the collection to provide standardised and high-quality judgments. In this case, all assessors have worked before in cultural heritage contexts and entities in conflict [17,27]. Also, the criteria to decide the relevance of the posts were made a priori before the assessment process.…”

Section: Relevance Judgmentmentioning

confidence: 99%

Building Cultural Heritage Reference Collections from Social Media through Pooling Strategies: The Case of 2020’s Tensions Over Race and Heritage

Otero

Martín-Rodilla

Parapar

2021

J. Comput. Cult. Herit.

View full text Add to dashboard Cite

Social networks constitute a valuable source for documenting heritage constitution processes or obtaining a real-time snapshot of a cultural heritage research topic. Many heritage researchers use social networks as a social thermometer to study these processes, creating, for this purpose, collections that constitute born-digital archives potentially reusable, searchable, and of interest to other researchers or citizens. However, retrieval and archiving techniques used in social networks within heritage studies are still semi-manual, being a time-consuming task and hindering the reproducibility, evaluation, and open-up of the collections created. By combining Information Retrieval strategies with emerging archival techniques, some of these weaknesses can be left behind. Specifically, pooling is a well-known Information Retrieval method to extract a sample of documents from an entire document set (posts in case of social network’s information), obtaining the most complete and unbiased set of relevant documents on a given topic. Using this approach, researchers could create a reference collection while avoiding annotating the entire corpus of documents or posts retrieved. This is especially useful in social media due to the large number of topics treated by the same user or in the same thread or post. We present a platform for applying pooling strategies combined with expert judgment to create cultural heritage reference collections from social networks in a customisable, reproducible, documented, and shareable way. The platform is validated by building a reference collection from a social network about the recent attacks on patrimonial entities motivated by anti-racist protests. This reference collection and the results obtained from its preliminary study are available for use. This real application has allowed us to validate the platform and the pooling strategies for creating reference collections in heritage studies from social networks.

show abstract

“…In the article "Assisting Forensic Identification through Unsupervised Information Extraction of Free Text Autopsy Reports: The Disappearances Cases during the Brazilian Military Dictatorship" [3], the authors analyze over 3000 reports of missing persons in São Paulo city during the Brazilian dictatorship. For this purpose, they make use of unsupervised methods of information extraction in Portuguese that let them identify named entities and terminology.…”

mentioning

confidence: 99%