A genetic algorithm for segmentation and information retrieval of SEC regulatory filings

ABSTRACT

A principal mechanism by which the SEC fulfills its missions of investor protection and market efficiency is the widespread dissemination of the information that publicly traded firms submit for disclosure. The continuing evolution of reporting standards like the International Financial Reporting Standards (IFRS) and the global convergence on XBRL as a syntax for sharing data address the quantitative dimension of reporting. This work complements the ongoing research on financial disclosure by helping investors learn from the textual, narrative portions of the filing. Our objective is to automatically segment SEC 10-K financial regulatory filings to facilitate structured retrieval and querying. In structured retrieval, terms are differentially weighted based upon the document segments in which a term appears. We leverage the regulatory instructions provided by the SEC to identify a set of semantic labels such as "Legal Proceedings" or "Management's Discussion and Analysis" that segment a 10-K annual report. We frame the problem of document segmentation as a search for semantic labels and use a genetic algorithm to segment each filing. We evaluate the genetic algorithm on a test set of 112 randomly selected regulatory filings and compare those results to a simple, greedy approach for information extraction and segmentation.

REFERENCES

• 1: Eugene Agichtein , Venkatesh Ganti, Mining reference tables for automatic text segmentation, Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining, August 22-25, 2004, Seattle, WA, USA [doi>10.1145/1014052.1014058]
• 2: Ricardo Baeza-Yates , Gonzalo Navarro, Integrating contents and structure in text retrieval, ACM SIGMOD Record, v.25 n.1, p.67-79, March 1996 [doi>10.1145/381854.381890]
• 3: Banko, M., Brill, E., Dumais, S. and Lin, J. AskMSR: Question Answering Using the Worldwide Web. AAAI Spring Symposium on Mining Answers from Texts and Knowledge Bases, 2002.
• 4: Vinayak Borkar , Kaustubh Deshmukh , Sunita Sarawagi, Automatic segmentation of text into structured records, ACM SIGMOD Record, v.30 n.2, p.175-186, June 2001
• 5: Mary Elaine Califf , Raymond J. Mooney, Relational learning of pattern-match rules for information extraction, Proceedings of the sixteenth national conference on Artificial intelligence and the eleventh Innovative applications of artificial intelligence conference innovative applications of artificial intelligence, p.328-334, July 18-22, 1999, Orlando, Florida, United States
• 6: U.S. EPA, Accidental Release Information Program (ARIP), 1999.
• 7: Weiguo Fan , Michael D. Gordon , Praveen Pathak, Genetic Programming-Based Discovery of Ranking Functions for Effective Web Search, Journal of Management Information Systems, v.21 n.4, p.37-56, Number 4/Spring 2005
• 8: Dayne Freitag, Information extraction from HTML: application of a general machine learning approach, Proceedings of the fifteenth national/tenth conference on Artificial intelligence/Innovative applications of artificial intelligence, p.517-523, July 1998, Madison, Wisconsin, United States
• 9: M. Gordon, Probabilistic and genetic algorithms in document retrieval, Communications of the ACM, v.31 n.10, p.1208-1218, Oct. 1988 [doi>10.1145/63039.63044]
• 10: John E. Hopcroft , Rajeev Motwani , Jeffrey D. Ullman, Introduction to Automata Theory, Languages, and Computation (3rd Edition), Addison-Wesley Longman Publishing Co., Inc., Boston, MA, 2006
• 11: Hsu, C.-N. and Chang, C.-C. Finite-State Transducers for Semi-Structured Text Mining. IJCAI Workshop on Text Mining: Foundations, Techniques and Application, 1999.
• 12: Kushmerick, N., Weld, D. S. and Doorenbos, R. Wrapper Induction for Information Extraction. IJCAI, 1997.
• 13: Lee, T. Using Regulatory Instructions for Information Extraction. AAAI Workshop on IIWeb, 2007.
• 14: Andrew McCallum , Dayne Freitag , Fernando C. N. Pereira, Maximum Entropy Markov Models for Information Extraction and Segmentation, Proceedings of the Seventeenth International Conference on Machine Learning, p.591-598, June 29-July 02, 2000
• 15: Ion Muslea , Steven Minton , Craig A. Knoblock, Hierarchical Wrapper Induction for Semistructured Information Sources, Autonomous Agents and Multi-Agent Systems, v.4 n.1-2, p.93-114, March-June 2001 [doi>10.1023/A:1010022931168]
• 16: Pathak, P., Gordon, M. D. and Fan, W. Effective Information Retrieval Using Genetic Algorithms Based Matching Functions Adaptation. HICSS, 2000.
• 17: Rylander, B. and Foster, J. Computational Complexity and Genetic Algorithms. Conference on Soft Computing, Advances in Fuzzy Systems and Evolutionary Computation, 2001.
• 18: U.S. SEC The Investor's Advocate: How the SEC Protects Investors, Maintains Market Integrity, and Facilitates Capital Formation. 2007.
• 19: U.S. SEC, EDGAR Full-Text Search FAQ. 2007.
• 20: U.S. SEC, Annual Report Pursuant to Section 13 or 15(d) (Form 10-K) General Instructions. 2005.
• 21: U.S. SEC, FAQ: XBRL Voluntary Filing Program. 2006.
• 22: Soderland, S. Learning to Extract Text-based Information from the World Wide Web. KDD, 1997.
• 23: White, J. W. Drilling for Disclosure: The Powerful Tool of Interactive Data. AAPG/SPE International Multidisciplinary Reserves Conference (Washington, DC, June 25, 2007).
• 24: Yang, J.-J., Korfhage, R. and Rasmussen, E. Query Improvement in Information Retrieval Using Genetic Algorithms. TREC, 1992.

INDEX TERMS

Primary Classification:
  J. Computer Applications
  J.1 ADMINISTRATIVE DATA PROCESSING

Additional Classification:
  H. Information Systems
  H.3 INFORMATION STORAGE AND RETRIEVAL
      H.3.5 On-line Information Services
          Subjects: Web-based services

General Terms:
Documentation, Management

Keywords:
SEC, document segmentation, genetic algorithms, structured information retrieval