Comparing manual text patterns and machine learning for classification of e-mails for automatic answering by a government agency

Hercules Dalianis; Jonas Sjöbergh; Eriks Sneiders

doi:10.1007/978-3-642-19437-5_19

Comparing manual text patterns and machine learning for classification of e-mails for automatic answering by a government agency

Hercules Dalianis^*
, Jonas Sjöbergh
, Eriks Sneiders

^*Corresponding author for this work

Stockholm University
KTH Royal Institute of Technology

Research output: Chapter in Book/Report/Conference proceeding › Conference paper › Research › peer-review

10 Citations (Scopus)

Abstract

E-mails to government institutions as well as to large companies may contain a large proportion of queries that can be answered in a uniform way. We analysed and manually annotated 4,404 e-mails from citizens to the Swedish Social Insurance Agency, and compared two methods for detecting answerable e-mails: manually-created text patterns (rule-based) and machine learning-based methods. We found that the text pattern-based method gave much higher precision at 89 percent than the machine learning-based method that gave only 63 percent precision. The recall was slightly higher (66 percent) for the machine learning-based methods than for the text patterns (47 percent). We also found that 23 percent of the total e-mail flow was processed by the automatic e-mail answering system.

Original language	English
Title of host publication	Computational Linguistics and Intelligent Text Processing - 12th International Conference, CICLing 2011, Proceedings
Publisher	Springer Verlag
Pages	234-243
Number of pages	10
Edition	PART 2
ISBN (Print)	9783642194368
DOIs	https://doi.org/10.1007/978-3-642-19437-5_19
Publication status	Published - 2011
Externally published	Yes

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Number	PART 2
Volume	6609 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 16 Peace, Justice and Strong Institutions

Keywords

automatic e-mail answering
E-government
machine learning
Naïve Bayes
SVM
text pattern matching

Access to Document

10.1007/978-3-642-19437-5_19

Cite this

Dalianis, H., Sjöbergh, J., & Sneiders, E. (2011). Comparing manual text patterns and machine learning for classification of e-mails for automatic answering by a government agency. In Computational Linguistics and Intelligent Text Processing - 12th International Conference, CICLing 2011, Proceedings (PART 2 ed., pp. 234-243). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6609 LNCS, No. PART 2). Springer Verlag. https://doi.org/10.1007/978-3-642-19437-5_19

Dalianis, Hercules ; Sjöbergh, Jonas ; Sneiders, Eriks. / Comparing manual text patterns and machine learning for classification of e-mails for automatic answering by a government agency. Computational Linguistics and Intelligent Text Processing - 12th International Conference, CICLing 2011, Proceedings. PART 2. ed. Springer Verlag, 2011. pp. 234-243 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2).

@inproceedings{7216899a980c41589ddded6481202bd6,

title = "Comparing manual text patterns and machine learning for classification of e-mails for automatic answering by a government agency",

abstract = "E-mails to government institutions as well as to large companies may contain a large proportion of queries that can be answered in a uniform way. We analysed and manually annotated 4,404 e-mails from citizens to the Swedish Social Insurance Agency, and compared two methods for detecting answerable e-mails: manually-created text patterns (rule-based) and machine learning-based methods. We found that the text pattern-based method gave much higher precision at 89 percent than the machine learning-based method that gave only 63 percent precision. The recall was slightly higher (66 percent) for the machine learning-based methods than for the text patterns (47 percent). We also found that 23 percent of the total e-mail flow was processed by the automatic e-mail answering system.",

keywords = "automatic e-mail answering, E-government, machine learning, Na{\"i}ve Bayes, SVM, text pattern matching",

author = "Hercules Dalianis and Jonas Sj{\"o}bergh and Eriks Sneiders",

year = "2011",

doi = "10.1007/978-3-642-19437-5\_19",

language = "English",

isbn = "9783642194368",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

number = "PART 2",

pages = "234--243",

booktitle = "Computational Linguistics and Intelligent Text Processing - 12th International Conference, CICLing 2011, Proceedings",

address = "Germany",

edition = "PART 2",

}

Dalianis, H, Sjöbergh, J & Sneiders, E 2011, Comparing manual text patterns and machine learning for classification of e-mails for automatic answering by a government agency. in Computational Linguistics and Intelligent Text Processing - 12th International Conference, CICLing 2011, Proceedings. PART 2 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 2, vol. 6609 LNCS, Springer Verlag, pp. 234-243. https://doi.org/10.1007/978-3-642-19437-5_19

Comparing manual text patterns and machine learning for classification of e-mails for automatic answering by a government agency. / Dalianis, Hercules; Sjöbergh, Jonas; Sneiders, Eriks.
Computational Linguistics and Intelligent Text Processing - 12th International Conference, CICLing 2011, Proceedings. PART 2. ed. Springer Verlag, 2011. p. 234-243 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6609 LNCS, No. PART 2).

Research output: Chapter in Book/Report/Conference proceeding › Conference paper › Research › peer-review

TY - GEN

T1 - Comparing manual text patterns and machine learning for classification of e-mails for automatic answering by a government agency

AU - Dalianis, Hercules

AU - Sjöbergh, Jonas

AU - Sneiders, Eriks

PY - 2011

Y1 - 2011

N2 - E-mails to government institutions as well as to large companies may contain a large proportion of queries that can be answered in a uniform way. We analysed and manually annotated 4,404 e-mails from citizens to the Swedish Social Insurance Agency, and compared two methods for detecting answerable e-mails: manually-created text patterns (rule-based) and machine learning-based methods. We found that the text pattern-based method gave much higher precision at 89 percent than the machine learning-based method that gave only 63 percent precision. The recall was slightly higher (66 percent) for the machine learning-based methods than for the text patterns (47 percent). We also found that 23 percent of the total e-mail flow was processed by the automatic e-mail answering system.

AB - E-mails to government institutions as well as to large companies may contain a large proportion of queries that can be answered in a uniform way. We analysed and manually annotated 4,404 e-mails from citizens to the Swedish Social Insurance Agency, and compared two methods for detecting answerable e-mails: manually-created text patterns (rule-based) and machine learning-based methods. We found that the text pattern-based method gave much higher precision at 89 percent than the machine learning-based method that gave only 63 percent precision. The recall was slightly higher (66 percent) for the machine learning-based methods than for the text patterns (47 percent). We also found that 23 percent of the total e-mail flow was processed by the automatic e-mail answering system.

KW - automatic e-mail answering

KW - E-government

KW - machine learning

KW - Naïve Bayes

KW - SVM

KW - text pattern matching

UR - https://www.scopus.com/pages/publications/79952274522

U2 - 10.1007/978-3-642-19437-5_19

DO - 10.1007/978-3-642-19437-5_19

M3 - Conference paper

AN - SCOPUS:79952274522

SN - 9783642194368

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 234

EP - 243

BT - Computational Linguistics and Intelligent Text Processing - 12th International Conference, CICLing 2011, Proceedings

PB - Springer Verlag

ER -

Dalianis H, Sjöbergh J, Sneiders E. Comparing manual text patterns and machine learning for classification of e-mails for automatic answering by a government agency. In Computational Linguistics and Intelligent Text Processing - 12th International Conference, CICLing 2011, Proceedings. PART 2 ed. Springer Verlag. 2011. p. 234-243. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2). doi: 10.1007/978-3-642-19437-5_19

Comparing manual text patterns and machine learning for classification of e-mails for automatic answering by a government agency

Abstract

Publication series

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this