by Sean O'Riain, Edward Curry, Robert Pinsker
Abstract:
Competitive business analysis is a manually intensive process that depends on analyst heuristics to gather and interpret relevant information from sources such as the SEC filings. The continuous growth in the volume of reports, and difficulty experienced with formats such as the eXtensible Business Reporting Language (XBRL) when sourcing information, challenge the viability of the analyst to conduct manual analysis. In particular analysis of the free text discussion sections of financial reports for qualitative data requires considerable effort, making comparison of qualitative data alternatives difficult. There is a clear need for an automatic linguistic analysis capability that can add meaning, structure and provide relevant contextual information to reduce the effort needed to analyse free text discussion sections. This paper introduces a qualitative interactive Decision Support System (QDSS), driven by an Ontology Based Linguistic Analysis Pipeline that leverages domain knowledge to drive linguistic analysis and generate structured, semantically marked-up data. The approach is evaluated within a competitive intelligence scenario where an analysis using a QDSS based on the pipeline output is compared against a manually conducted analysis. Experiment results report a 37% performance improvement in finding relevant information and usability results, on the pipelines contribution to competitive analysis task structuring and information provision.
Reference:
Sean O'Riain, Edward Curry, Robert Pinsker, "Competitive Analysis of Business Filings Using Ontologies and Linguistic Analysis", In Joint SIGDSS & TUN Business Intelligence Congress 3: Driving Innovation through Big Data Analytics, Orlando, Florida, pp. 1-15, 2012.
Bibtex Entry:
@inproceedings{Sean2012,
abstract = {Competitive business analysis is a manually intensive process that depends on analyst heuristics to gather and interpret relevant information from sources such as the SEC filings. The continuous growth in the volume of reports, and difficulty experienced with formats such as the eXtensible Business Reporting Language (XBRL) when sourcing information, challenge the viability of the analyst to conduct manual analysis. In particular analysis of the free text discussion sections of financial reports for qualitative data requires considerable effort, making comparison of qualitative data alternatives difficult. There is a clear need for an automatic linguistic analysis capability that can add meaning, structure and provide relevant contextual information to reduce the effort needed to analyse free text discussion sections. This paper introduces a qualitative interactive Decision Support System (QDSS), driven by an Ontology Based Linguistic Analysis Pipeline that leverages domain knowledge to drive linguistic analysis and generate structured, semantically marked-up data. The approach is evaluated within a competitive intelligence scenario where an analysis using a QDSS based on the pipeline output is compared against a manually conducted analysis. Experiment results report a 37% performance improvement in finding relevant information and usability results, on the pipelines contribution to competitive analysis task structuring and information provision.},
address = {Orlando, Florida},
author = {O'Riain, Sean and Curry, Edward and Pinsker, Robert},
booktitle = {Joint SIGDSS & TUN Business Intelligence Congress 3: Driving Innovation through Big Data Analytics},
file = {:Users/ed/Library/Application Support/Mendeley Desktop/Downloaded/O'Riain, Curry, Pinsker - 2012 - Competitive Analysis of Business Filings Using Ontologies and Linguistic Analysis.pdf:pdf},
keywords = {Competitive analysis,business intelligence,information extraction,linguistic analysis,ontology based information extraction,ontology development,semantic technologies},
pages = {1--15},
title = {{Competitive Analysis of Business Filings Using Ontologies and Linguistic Analysis}},
url = {http://www.edwardcurry.org/publications/SOR-CompetitiveAnalysisOfBusinessFilings.pdf},
year = {2012}
}