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

^*Šī darba korespondējošais autors

Stockholm University
KTH Royal Institute of Technology

Zinātniskās darbības rezultāts: Nodaļa grāmatā/enciklopēdijā/konferences krājumā › Konferences zinātniskais raksts › Pētniecība › koleģiāli recenzēts

10 Atsauces (Scopus)

Kopsavilkums

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.

Oriģinālvaloda	Angļu
Rīkotāja publikācijas nosaukums	Computational Linguistics and Intelligent Text Processing - 12th International Conference, CICLing 2011, Proceedings
Izdevējs	Springer Verlag
Lapas	234-243
Lapu skaits	10
Izdevums	PART 2
ISBN (Drukātā versija)	9783642194368
DOIs	https://doi.org/10.1007/978-3-642-19437-5_19
Publikācijas statuss	Publicēts - 2011
Ārēji publicēts	Jā

Publikāciju sērijas

Nosaukums	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Numurs	PART 2
Sējums	6609 LNCS
ISSN (Drukātā versija)	0302-9743
ISSN (Elektroniskā versija)	1611-3349

ANO IAM

Šis izpildes rezultāts palīdz sasniegt šādus ANO ilgtspējīgas attīstības mērķus (IAM)

16. IAM — Miers, Taisnīgums un Spēcīgas Iestādes

Piekļuve Dokumentiem

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

Citi faili un saites

Saite uz publikāciju Scopus

Citēt šo

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. (gadā) Computational Linguistics and Intelligent Text Processing - 12th International Conference, CICLing 2011, Proceedings (PART 2 red., lpp. 234-243). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Sēj. 6609 LNCS, Nr. 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. red. Springer Verlag, 2011. lpp. 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. (kur) Computational Linguistics and Intelligent Text Processing - 12th International Conference, CICLing 2011, Proceedings. PART 2 red., Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Nr. PART 2, sēj. 6609 LNCS, Springer Verlag, lpp. 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. red. Springer Verlag, 2011. lpp. 234-243 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Sēj. 6609 LNCS, Nr. PART 2).

Zinātniskās darbības rezultāts: Nodaļa grāmatā/enciklopēdijā/konferences krājumā › Konferences zinātniskais raksts › Pētniecība › koleģiāli recenzēts

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. (gadā) Computational Linguistics and Intelligent Text Processing - 12th International Conference, CICLing 2011, Proceedings. PART 2 red. Springer Verlag. 2011. lpp. 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

Kopsavilkums

Publikāciju sērijas

ANO IAM

Piekļuve Dokumentiem

Citi faili un saites

Nospiedums

Citēt šo