• Contenu principal
  • Menu
OpenEdition Books
  • Accueil
  • Catalogue de 16000 livres
  • Éditeurs
  • Auteurs
  • Facebook
  • X
  • Partager
    • Facebook

    • X

    • Accueil
    • Catalogue de 16000 livres
    • Éditeurs
    • Auteurs
  • Ressources numériques en sciences humaines et sociales

    • OpenEdition
  • Nos plateformes

    • OpenEdition Books
    • OpenEdition Journals
    • Hypothèses
    • Calenda
  • Bibliothèques

    • OpenEdition Freemium
  • Suivez-nous

  • Lettre d’information
OpenEdition Search

Redirection vers OpenEdition Search.

À quel endroit ?
  • Accademia University Press
  • ›
  • Collana dell'Associazione Italiana di Li...
  • ›
  • Proceedings of the Seventh Italian Confe...
  • ›
  • Contributed Papers
  • ›
  • A Resource for Detecting Misspellings an...
  • Accademia University Press
  • Accademia University Press
    Accademia University Press
    Informations sur la couverture
    Table des matières
    Liens vers le livre
    Informations sur la couverture
    Table des matières
    Formats de lecture

    Plan

    Plan détaillé Texte intégral 1. Introduction 2. Related Work 3. Data Analysis 4. Dictionaries Creation and Evaluation 6. Conclusions and Future Work Bibliographie Notes de bas de page Auteurs

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Ce livre est recensé par

    Précédent Suivant
    Table des matières

    A Resource for Detecting Misspellings and Denoising Medical Text Data

    Enrico Mensa, Gian Manuel Marino, Davide Colla, Matteo Delsanto et Daniele P. Radicioni

    p. 296-302

    Résumés

    In this paper we propose a method for collecting a dictionary to deal with noisy medical text documents. The quality of such Italian Emergency Room Reports is so poor that in most cases these can be hardly automatically elaborated; this also holds for other languages (e.g., English), with the notable difference that no Italian dictionary has been proposed to deal with this jargon. In this work we introduce and evaluate a resource designed to fill this gap.1

    In questo lavoro illustriamo un metodo per la costruzione di un dizionario dedicato all’elaborazione di documenti medici, la porzione delle cartelle cliniche annotata nei reparti di pronto soccorso. Questo tipo di documenti è così rumoroso che in genere le cartelle cliniche difficilmente posono essere direttamente elaborate in maniera automatica. Pur essendo il problema di ripulire questo tipo di documenti un problema rilevante e diffuso, non esisteva un dizionario completo per trattare questo linguaggio settoriale. In questo lavoro proponiamo e valutiamo una risorsa finalizzata a condurre questo tipo di elaborazione sulle cartelle cliniche.

    Remerciements

    The first author was supported by a grant provided by Università degli Studi di Torino. This research is also supported by Fondazione CRT, RF 2019.2263.

    Texte intégral Bibliographie Notes de bas de page Auteurs

    Texte intégral

    1. Introduction

    1Noise in textual data is a very common phenomenon afflicting text documents, especially when dealing with informal texts such as chats, SMS and e-mails. This kind of text inherently contains spelling errors, special characters, non-standard word forms, grammar mistakes, and so on (Liu, Weng, and Jiang 2012). In this work we focus on a type of text which can also be very noisy: emergency room reports. In the broader frame of a project aimed at detecting injuries stemming from violence acts in narrative texts contained in emergency room reports, we recently developed the ViDeS, so dubbed after ‘Violence Detection System’ (Mensa et al. 2020). This system is concerned with categorizing textual descriptions as containing violence-related injuries (V) vs. non-violence-related injuries (NV), which is a relevant task to the ends of devising alerting mechanisms to track and prevent violence episodes. ViDeS combines a neural architecture which performs the categorization step (thus discriminating V and NV records) and a Framenet-based approach, whereby semantic roles are represented through a synthetic description employing a set of word embeddings.2 More specifically, a model of violent event has been devised: records that are recognized as containing violence-related injuries are further processed by an explanation module, which is charged to individuate the main elements corroborating that categorization (V) by identifying the involved agent, the type of injury, the involved body district etc.. Explaining the categorization ultimately involves filling the semantic components of the violence frame.

    2All such elements contribute to recognizing a violent event as the source of the injuries complained by ER patients.

    3During the development of ViDeS we realized that in order to run sophisticated algorithms for the detection and extraction of such violent traits we needed to cope with the noise contained in the input medical records. Some efforts have been invested to deal with different sorts of linguistic phenomena menacing the comprehension of texts; however, most existing works are focused on the English language, and rely on dictionaries that cannot be directly employed on Italian text documents.

    4In this preliminary work we start to tackle the issue of noisy words in medical records for Italian texts, by specifically focusing on misspellings. Our contribution is twofold: we first manually explore the dataset by analyzing a small sample of records in order to determine whether the main traits and issues present in other languages are also shared by Italian reports; secondly, we collect, merge and evaluate a set of Italian dictionaries, which constitute a brick fundamental to build any domain specific spell-checking algorithm (López-Hernández, Almela, and Valencia-Garcı́a 2019).

    2. Related Work

    5Literature shows a limited but significant interest on the issue of detecting and correcting noisy medical text documents; nonetheless, some commonalities underlying this sort of text can be drawn.

    6Medical texts are often very noisy; among the most common mistakes we mention mistyping, lack or improper use of punctuation, grammatical errors and domain-specific abbreviations and Latin medical terminology (Siklósi, Novák, and Prószéky 2013). This is mainly due to the nature of the records themselves, and to the fact that the medical personnel compiling the entries is often under pressure and in a hurry.

    7Most of the spelling correction approaches have been carried out for English, with the exception of research in Swedish (Dziadek, Henriksson, and Duneld 2017) and Hungarian (Siklósi, Novák, and Prószéky 2013), while no work has been found dealing with the Italian language. Regarding the methodologies, most works focus on non-word errors, while disregarding grammatical and real word mistakes. Non-word mistakes occur when a misspelling error produces a word that does not exist, such as ‘patienz’ instead of ’patient’, while real word mistakes occur when a word is mistakenly replaced with another –existing– one, like the substitution of ‘abuse’ with ‘amuse’. The adopted algorithms are diverse, with the prevalence of approaches relying on embeddings (Kilicoglu et al. 2015; Workman et al. 2019) or regular expressions and rule-based systems (Patrick et al. 2010; Sayle, Xie, and Muresan 2012; Lai et al. 2015). However, basically all contributions adopt a preliminary dictionary look-up step (López-Hernández, Almela, and Valencia-Garcı́a 2019). To this purpose, besides the general dictionaries provided in toolkits such as Aspell and Google Spell Checker,3 authors often rely on (medical) domain-specific dictionaries, such as The Unified Medical Language System (UMLS) (Aoki et al. 2004), the Systematized Nomenclature of Medicine-Clinical Terms (SNOMED-CT 2020) and The SPECIALIST Lexicon (Browne, McCray, and Srinivasan 2000). It is thus evident that the development of analogous resources for the Italian language is a crucial step for the design of tools and systems aimed at dealing with the spell-checking of Italian medical text documents.

    8Besides the treatment of misspellings, there are also works specifically focused on abbreviations. For instance, in (Wu et al. 2011) the authors present a corpus-based method to create a lexical resource of English clinical abbreviations via several machine learning algorithms. The resource has been used to automatically detect and expand abbreviations, and obtained interesting experimental results. More recently, another approach proposed in (Kreuzthaler et al. 2016) focuses on abbreviations ending with a period character; the proposed technique puts together statistical and dictionary-based strategies to detect abbreviations in German clinical narratives.

    9In the present work we are not proposing a specific technique for dealing with abbreviations, we are rather concerned with misspellings. However, the approaches already proposed for other languages will be considered in future work to also treat Italian abbreviations in our dataset.

    3. Data Analysis

    10We analyze real data coming from a set of emergency room reports collected in Italian hospitals by the Italian National Institute of Health in the frame of the SINIACA project (Pitidis et al. 2014). The SINIACA project, so dubbed after ‘Sistema Informativo Nazionale sugli Incidenti in Ambiente di Civile Abitazione’ (National Information System on Accidents in Civil Housing Environment), is the Italian branch of the European Injury Database (EU-IDB) (Lyons, Kisse, and Rogmans 2015), an EU-wide surveillance system concerned with accidents, collecting data from hospital emergency departments according to the EU recommendation no. C 164/2007/01, aimed at injury prevention and safety promotion.

    Table 1

    Complete

    Sample

    Number of entries

    136,144

    592

    Number of tokens

    2, 329, 840

    14, 137

    Number of unique tokens

    49, 116

    1, 842

    Avg tokens per record

    17.11

    23.88

    Figures describing the complete dataset and the sample selected for manual annotation.

    Dataset

    11The whole dataset amounts to 136,144 non-empty entries, 592 of which were randomly selected for the manual analysis. Table 1 reports some figures describing the dataset. Double spaces and punctuation redundancy have been fixed through regular expressions, while tokens have been extracted by splitting the sentences based on spaces. Also, tokens containing numbers are presently discarded.

    Analysis result

    12We performed a manual analysis on the subset of the original dataset: the 592 randomly selected entries herein were manually examined, and for each entry we looked for noisy words. Three main types of words were annotated: i) misspellings: a wrongly typed word, e.g., fratura instead of frattura – fracture; ii) abbreviations: a shortened form of a word or phrase, e.g., dx instead of destra – right; iii) acronyms: a word formed from the initial letters of other words, e.g., ps instead of pronto soccorso – emergency room. Interestingly enough, both abbreviations and acronyms can be at least partly considered as domain dependent: for example, in different settings, ps may denote post scriptum (something added at a later time, likely a letter, after the signature), but also ‘Polizia di Stato’ (Police) or ‘previdenza sociale’ (social security). Dealing with such phenomena thus involves accessing a context dependent knowledge base that allows selecting the utterance appropriate for the context at hand. We are presently concerned with misspellings, acronyms and abbreviations as noise, but only the first category can be actually considered as an error. More specifically, while misspellings are actual errors, abbreviations and acronyms belong to a domain-specific language, and these are way too specific to be recognized as legitimate words through a general-purpose dictionary. As seen in literature, misspellings and abbreviations/acronyms must be treated with different techniques, and in this work we mainly focus on tackling the first category, while also obtaining interesting insights regarding the second one.

    13Table 2 illustrates the results of the annotation process. We discovered that the dataset contains a lot of noise, amounting to almost the 10% of the tokens, on average 2 noisy tokens per record. By looking separately at the different typologies of noise we observe that misspellings are more scattered and diverse, while the usage of abbreviations and acronyms seems to be more coherent: we have 670 instances of abbreviations but only 76 unique abbreviations, while 304 out of the 433 instances of misspellings are unique. This phenomenon is also depicted in Figure 1, where we provide the log-log plot of the frequency of each misspelling, abbreviation and acronym ordered by rank. We observe that the distribution of abbreviations and acronyms has a different magnitude, but is very similar in shape; on the other side, the misspellings are clearly more scattered with a very long tail of items appearing only once.

    Table 2

    With repetitions

    Unique

    Average

    Noisy tokens

    1, 336 (9.4%)

    424 (3%)

    2.25

    Misspellings

    433 (3%)

    304 (2%)

    0.73

    Abbreviations

    670 (4.7%)

    76 (0.5%)

    1.13

    Acronyms

    233 (1.6%)

    45 (0.3%)

    0.39

    Noise distribution on the annotated dataset; between parenthesis we report the per-centage over the total number of tokens, while the last column indicates the average per record.

    Figure 1

    Image 10000000000003B0000002C3529E2554C37FE923.jpg

    Log-log plot showing the frequency of misspellings, abbreviations and acronyms over the annotated dataset. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

    4. Dictionaries Creation and Evaluation

    14The manual analysis uncovered characteristics and features that are in line with those found in literature for English datasets (López-Hernández, Almela, and Valencia-Garcı́a 2019). However, to allow the development of spell-checkers for Italian medical texts, another key component is still missing: most approaches aimed at error detection rely on dictionaries to determine if a token is a legitimate word or not. In fact, the simplest implementation of misspellings detection is as follows: if we have at our disposal the set W containing all of the terms of a given language, joined to all terms pertaining the specific domain at hand, any word Image 1000000000000037000000130B041788BC847E8A.jpgcan be likely considered as a misspell. To the best of our knowledge, no such dictionary exists that is able to cope with Italian medical text documents, so we built a resource to answer to this need.

    4.1 Source Dictionaries

    15The automatic development of a dictionary is not a trivial task. We want to reach the highest possible coverage for both general terms and specific medical terminology, but at the same time we cannot rely too much on unverified sources (e.g., crowd-sourced data) with the risk of introducing misspellings and errors into the dictionary. We selected different sources and arranged them into four main classes:

    • MED: a collection of medical terms built by putting together five medical online dictionaries (torrinomedica.it 2020; abcsalute.it 2020; codifa.it 2020; my-personaltrainer.it 2020a, 2020b), containing medical specific terms and medications names;

    • ITA: a collection of Italian terms built by merging three well-known Italian online dictionaries (Hoepli 2020; Sabatini-Colletti 2020; De Mauro 2020);

    • WMED: a collection of terms from Wikipedia pages pertaining the medical domain. The list of Wikipedia medical pages has been obtained by querying the SPARQL endpoint of Wikidata (Vrandečić and Krötzsch 2014), while the pages have been taken from the 20 August 2020 Wikipedia dump;

    • WMOV: since medical records also contain a brief narrative text of the events that led to the (either violent or accidental) injuries, we added terms associated to eventive and narrative genres by collecting Wikipedia pages pertaining to movies, television series and literary work that are expected to contain narrative terminology.

    16The set of terms extracted from Wikipedia can potentially contain misspellings and errors, and so we also set a frequency minimum which allows for the pruning of the tokens herein. We annotate this parameter with a subscript next to the set name, e.g., WMOV1 indicates that the threshold was set to 1 for the terms frequency.

    5. Evaluation

    Building the dataset

    17In order to assess the quality of the collected dictionaries we started from the 49,116 unique tokens in the dataset, removed the stop words4 and randomly selected 5,000 of them to be manually annotated. The annotation was carried out by four of the authors of this paper. The selection algorithm was designed so to increase the probability of a token to be selected in accordance to its frequency in the dataset. These 5,000 tokens were then annotated with one of the following four classes: correct words (regardless of their domain specificity), abbreviations, acronyms and misspellings. The first three classes represent terms that should be found in our resource, while the last category contains words that should not be present in the dictionary. Table 3 reports the statistics featuring the dataset annotated for evaluation purposes.

    Table 3

    Figures of the 5000 annotated tokens used to evaluate the dictionaries.

    Evaluating the dictionary

    18In Table 4 we report the results of the dictionaries evaluation. Each dictionary has been built by taking into consideration one or more of the previously presented sources. Multiple sources have been simply merged into a unique set of terms, without repetitions. We assess the quality of each dictionary via two measures, coverage and correctness. The coverage is the percentage of words that were found in the dictionary (either correct words, abbreviations or acronyms), while the correctness is the percentage of misspellings that were not present in the dictionary. We considered different combinations of the sources, the tuning of the frequency-based filtering parameter, and an additional lemmatization step.

    Table 4

    Terms

    Coverage

    Correctness

    F1-Score

    Correct

    Words

    Abbreviations

    Acronyms

    ITA

    124,

    494

    .542

    .980

    .700

    .573

    .179

    .206

    MED, ITA

    155,

    650

    .621

    .975

    .759

    .652

    .228

    .261

    MED, ITA, WMED0

    287,

    279

    .897

    .907

    .902

    .918

    .521

    .785

    MED, ITA, WMED0, WMOV0

    511,

    827

    .926

    .863

    .894

    .941

    .641

    .873

    MED, ITA, WMED1, WMOV1

    343,

    264

    .906

    .898

    .902

    .925

    .586

    .793

    MED, ITA, WMED1, WMOV5

    266,

    633

    .892

    .922

    .907

    .912

    .554

    .761

    (LEM) MED, ITA, WMED1, WMOV5

    227,

    895

    .903

    .896

    .900

    .926

    .559

    .674

    Results of the evaluation of the considered dictionaries. The first column reports the size of each dictionary, the second to fourth columns provide coverage and correctness along with their har-monic mean, while the last three columns illustrate the coverage of our dictionaries on tokens that were annotated as correct words, abbreviations and acronyms.

    19We observe that both the ITA and the MED sets are fundamentally correct, even though they also include words that in the common usage are frequently misspelled, such as passeggiero in place of the correct form passeggero. On the other side, its .62 coverage is unsatisfactory (please refer to the second row of Table 4, MED, ITA); it also witnesses that medical jargon is only partially grasped by dictionaries in the MED set. As expected, the introduction of terms from Wikipedia improves the coverage, but with detrimental effect on the correctness. This also holds for the WMOV set, which is rich but also pretty noisy. By fine tuning the frequency thresholds of both WMED and WMOV we were able to prune most of the noise and to preserve the coverage at the same time, finally obtaining a good dictionary with the combination MED, ITA, WMED1, WMOV5.

    20This setting was also tested by applying a lemmatization step on both Wikipedia terms and our dataset tokens. Interestingly, the lemmatization introduces more mistakes than it solves: this is due the the fact that unpredictably the lemmatizer converts misspellings into legitimate words that do not necessarily correspond to their correct spelling. This fact shows also that lemmatization, which is acknowledged as a task almost completely solved from a scientific point of view, still poses relevant issues for the medical jargon and for domain-specific languages more in general.

    21A lot of abbreviations are not yet covered in the dictionary. Once again, these abbreviations are dataset-specific (and perhaps also follow local uses rather than widely accepted practices), and thus these are very hard to find even on specialized public medical resources. For instance, incid (incidente – accident) appears very frequently and its easily understandable by humans but its not a common or medical abbreviation. The same phenomenon can also be observed on acronyms, that are less sparse and more adherent to widely accepted practices and standards.

    6. Conclusions and Future Work

    22In this work we tackled the issue of detecting textual noise in Italian room emergency reports, focusing specifically on misspellings. Firstly we examined the reports and found out that the sorts of issues reported in literature for other languages can also be found in Italian text documents. Secondly, we developed and evaluated an Italian dictionary suited for the task of noise detection. In future work we plan to expand the dictionary by including the terms from the Italian ICD-9 and ICD-10 (International Classification of Diseases), that may be useful to interpret acronyms and resolve abbreviations. Moreover, we plan to employ this dictionary in a fully fledged spell-checking system. Finally, the usage of semantic —sense indexed— representations such as, e.g., (Mensa, Radicioni, and Lieto 2018) and (Colla, Mensa, and Radicioni 2020a, 2020b) will be explored, in order to deal with real word mistakes, and more in general contextual information (Basile, Caselli, and Radicioni 2019) will be considered as a main cue in order to uncover and correct this sort of errors. For example, by leveraging the terminology surrounding noisy tokens we plan to distinguish the more scattered misspellings from the other terms that are not present in our dictionary.

    Bibliographie

    Des DOI sont automatiquement ajoutés aux références bibliographiques par Bilbo, l’outil d’annotation bibliographique d’OpenEdition. Ces références bibliographiques peuvent être téléchargées dans les formats APA, Chicago et MLA.

    Format

    • APA
    • Chicago
    • MLA
    Basile, V., Caselli, T., & Radicioni, D. P. (2019). Meaning in Context: Ontologically and linguistically motivated representations of objects and events (V. Basile, T. Caselli, & D. P. Radicioni, Eds.). IOS Press. https://doi.org/10.3233/ao-190221
    Gildea, D., & Jurafsky, D. (2002). Automatic Labeling of Semantic Roles. MIT Press - Journals. https://doi.org/10.1162/089120102760275983
    Proceedings of the 20th ACM conference on Hypertext and hypermedia. (2009). Presented at the HT ’09: 20th ACM Conference on Hypertext and Hypermedia. https://doi.org/10.1145/1557914
    Mensa, E., Colla, D., Dalmasso, M., Giustini, M., Mamo, C., Pitidis, A., & Radicioni, D. P. (2020). Violence detection explanation via semantic roles embeddings. Springer Science and Business Media LLC. https://doi.org/10.1186/s12911-020-01237-4
    Mensa, E., Radicioni, D. P., & Lieto, A. (2018). COVER: a linguistic resource combining common sense and lexicographic information. Springer Science and Business Media LLC. https://doi.org/10.1007/s10579-018-9417-z
    Vrandečić, D., & Krötzsch, M. (2014). Wikidata. Association for Computing Machinery (ACM). https://doi.org/10.1145/2629489
    Basile, Valerio, Tommaso Caselli, and Daniele P. Radicioni. “Meaning in Context: Ontologically and Linguistically Motivated Representations of Objects and Events”. Edited by Valerio Basile, Tommaso Caselli, and Daniele P. Radicioni. Applied Ontology. IOS Press, November 12, 2019. doi:10.3233/ao-190221.
    Gildea, Daniel, and Daniel Jurafsky. “Automatic Labeling of Semantic Roles”. Computational Linguistics. MIT Press - Journals, September 2002. doi:10.1162/089120102760275983.
    “Proceedings of the 20th ACM Conference on Hypertext and Hypermedia”. []. ACM, June 29, 2009. doi:10.1145/1557914.
    Mensa, Enrico, Davide Colla, Marco Dalmasso, Marco Giustini, Carlo Mamo, Alessio Pitidis, and Daniele P. Radicioni. “Violence Detection Explanation via Semantic Roles Embeddings”. BMC Medical Informatics and Decision Making. Springer Science and Business Media LLC, October 15, 2020. doi:10.1186/s12911-020-01237-4.
    Mensa, Enrico, Daniele P. Radicioni, and Antonio Lieto. “COVER: A Linguistic Resource Combining Common Sense and Lexicographic Information”. Language Resources and Evaluation. Springer Science and Business Media LLC, June 21, 2018. doi:10.1007/s10579-018-9417-z.
    Vrandečić, Denny, and Markus Krötzsch. “Wikidata”. Communications of the ACM. Association for Computing Machinery (ACM), September 23, 2014. doi:10.1145/2629489.
    Basile, Valerio, et al. “Meaning in Context: Ontologically and Linguistically Motivated Representations of Objects and Events”. Applied Ontology, edited by Valerio Basile et al., vols. 14, nos. 4, IOS Press, 12 Nov. 2019, pp. 335-41. Crossref, https://doi.org/10.3233/ao-190221.
    Gildea, Daniel, and Daniel Jurafsky. “Automatic Labeling of Semantic Roles”. Computational Linguistics, vols. 28, nos. 3, MIT Press - Journals, Sept. 2002, pp. 245-88. Crossref, https://doi.org/10.1162/089120102760275983.
    Proceedings of the 20th ACM Conference on Hypertext and Hypermedia. [], ACM, 29 June 2009. Crossref, https://doi.org/10.1145/1557914.
    Mensa, Enrico, et al. “Violence Detection Explanation via Semantic Roles Embeddings”. BMC Medical Informatics and Decision Making, vols. 20, no. 1, Springer Science and Business Media LLC, 15 Oct. 2020. Crossref, https://doi.org/10.1186/s12911-020-01237-4.
    Mensa, Enrico, et al. “COVER: A Linguistic Resource Combining Common Sense and Lexicographic Information”. Language Resources and Evaluation, vols. 52, nos. 4, Springer Science and Business Media LLC, 21 June 2018, pp. 921-48. Crossref, https://doi.org/10.1007/s10579-018-9417-z.
    Vrandečić, Denny, and Markus Krötzsch. “Wikidata”. Communications of the ACM, vols. 57, nos. 10, Association for Computing Machinery (ACM), 23 Sept. 2014, pp. 78-85. Crossref, https://doi.org/10.1145/2629489.

    Cette bibliographie a été enrichie de toutes les références bibliographiques automatiquement générées par Bilbo en utilisant Crossref.

    abcsalute.it. 2020. “Abcsalute.it - Dizionario Medico.” http://www.abcsalute.it/dizionario-medico.

    Kiyoko F. Aoki, Atsuko Yamaguchi, Nobuhisa Ueda, Tatsuya Akutsu, Hiroshi Mamitsuka, Susumu Goto, and Minoru Kanehisa. 2004. “KCaM (Kegg Carbohydrate Matcher): A Software Tool for Analyzing the Structures of Carbohydrate Sugar Chains.” Nucleic Acids Research 32 (suppl_2): W267–W272.

    Valerio Basile, Tommaso Caselli, and Daniele P. Radicioni. 2019. “Meaning in Context: Ontologically and linguistically motivated representations of objects and events.” Applied Ontology 14: 335–41. https://doi.org/10.3233/AO-190221.

    10.3233/AO-190221 :

    Allen C. Browne, Alexa T McCray, and Suresh Srinivasan. 2000. “The Specialist Lexicon.” National Library of Medicine Technical Reports, 18–21.

    codifa.it. 2020. “codifa.it - Dizionario dei Farmaci.” https://www.codifa.it/farmaci.

    Davide Colla, Enrico Mensa, and Daniele P. Radicioni. 2020a. “LessLex: Linking Multilingual Embeddings to Sense Representations of Lexical Items.” Computational Linguistics 46 (2): 289–333. https://doi.org/10.1162/coli\_a\_00375.

    10.1162/coli\_a\_00375 :

    Davide Colla, Enrico Mensa, and Daniele P. Radicioni. 2020b. “Novel Metrics for Computing Semantic Similarity with Sense Embeddings.” Knowledge-Based Systems 206: 106346. https://direct.mit.edu/coli/article/46/2/289/93366/LessLex-Linking-Multilingual-Embeddings-to-SenSe.

    De Mauro. 2020. “Dizionario Italiano Nuovo de Mauro.” https://dizionario.internazionale.it/.

    Juliusz Dziadek, Aron Henriksson, and Martin Duneld. 2017. “Improving Terminology Mapping in Clinical Text with Context-Sensitive Spelling Correction.” Informatics for Health: Connected Citizen-Led Wellness and Population Health 235: 241.

    Davide Gianfelice, Leonardo Lesmo, Monica Palmirani, Daniele Perlo, and Daniele P. Radicioni. 2013. “Modificatory Provisions Detection: a Hybrid NLP Approach.” In Proceedings of ICAIL 2013: XIV International Conference on Artificial Intelligence and Law, edited by Bart Verheij, 43–52. ACM.

    Daniel Gildea and Daniel Jurafsky. 2002. “Automatic Labeling of Semantic Roles.” Computational Linguistics 28 (3): 245–88.

    10.1162/089120102760275983 :

    Hoepli. 2020. “Dizionario Italiano Hoepli.” https://dizionari.repubblica.it/italiano.html.

    Halil Kilicoglu, Marcelo Fiszman, Kirk Roberts, and Dina Demner-Fushman. 2015. “An Ensemble Method for Spelling Correction in Consumer Health Questions.” In AMIA Annual Symposium Proceedings, 2015:727. American Medical Informatics Association.

    Markus Kreuzthaler, Michel Oleynik, Alexander Avian, and Stefan Schulz. 2016. “Unsupervised Abbreviation Detection in Clinical Narratives.” In Proceedings of the Clinical Natural Language Processing Workshop (Clinicalnlp), 91–98.

    Kenneth H. Lai, Maxim Topaz, Foster R Goss, and Li Zhou. 2015. “Automated Misspelling Detection and Correction in Clinical Free-Text Records.” Journal of Biomedical Informatics 55: 188–95.

    Leonardo Lesmo, Alessandro Mazzei, and Daniele P. Radicioni. 2009. “Extracting Semantic Annotations from Legal Texts.” In Proceedings of the International Conference on Hypertext, Ht09, 167–72. Turin, Italy: ACM.

    10.1145/1557914 :

    Fei Liu, Fuliang Weng, and Xiao Jiang. 2012. “A Broad-Coverage Normalization System for Social Media Language.” In Proceedings of the 50th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 1035–44.

    Jésica López-Hernández, Ángela Almela, and Rafael Valencia-Garcı́a. 2019. “Automatic Spelling Detection and Correction in the Medical Domain: A Systematic Literature Review.” In International Conference on Technologies and Innovation, 95–108. Springer.

    Ronan Lyons, Rupert Kisse, and Wim Rogmans. 2015. “EU-Injury Database Introduction to the Functioning of the Injury Database (Idb).” European Association for Injury Prevention; Safety Promotion (EuroSafe).

    Enrico Mensa, Davide Colla, Marco Dalmasso, Marco Giustini, Carlo Mamo, Alessio Pitidis, and Daniele P. Radicioni. 2020. “Violence Detection Explanation via Semantic Roles Embeddings.” BMC Medical Informatics and Decision Making 20 (1): 263–75. https://doi.org/10.1186/s12911-020-01237-4.

    10.1186/s12911-020-01237-4 :

    Enrico Mensa, Daniele P. Radicioni, and Antonio Lieto. 2018. “COVER: A Linguistic Resource Combining Common Sense and Lexicographic Information.” Language Resources and Evaluation 52 (4): 921–48. https://doi.org/10.1007/s10579-018-9417-z.

    10.1007/s10579-018-9417-z :

    my-personaltrainer.it. 2020a. “Lista delle Malattie di My Personal Trainer.” https://www.my-personaltrainer.it/malattie_a_z.php.

    my-personaltrainer.it. 2020b. “Lista di Sintomi di My Personal Trainer.” https://www.my-personaltrainer.it/sintomi_a_z.php.

    Jon Patrick, Mojtaba Sabbagh, Suvir Jain, and Haifeng Zheng. 2010. “Spelling Correction in Clinical Notes with Emphasis on First Suggestion Accuracy.” In Proceedings of 2nd Workshop on Building and Evaluating Resources for Biomedical Text Mining (Biotxtm2010), 1–8.

    Alessio Pitidis, Gianni Fondi, Marco Giustini, Eloïse Longo, Giuseppe Balducci, Gruppo di lavoro SINIACA-IDB, and Dipartimento di Ambiente e Connessa Prevenzione Primaria, ISS. 2014. “Il Sistema SINIACA-IDB per la sorveglianza degli incidenti.” Notiziario Dell’Istituto Superiore Di Sanità 27 (2): 11–16.

    Sabatini-Colletti. 2020. “Dizionario Italiano Sabatibi Colletti.” https://dizionari.corriere.it/dizionario_italiano/.

    Roger Sayle, Paul Hongxing Xie, and Sorel Muresan. 2012. “Improved Chemical Text Mining of Patents with Infinite Dictionaries and Automatic Spelling Correction.” Journal of Chemical Information and Modeling 52 (1): 51–62.

    Borbála Siklósi, Attila Novák, and Gábor Prószéky. 2013. “Context-Aware Correction of Spelling Errors in Hungarian Medical Documents.” In International Conference on Statistical Language and Speech Processing, 248–59. Springer.

    SNOMED-CT. 2020. “International Health Terminology Standards Development Organisation.” http://www.ihtsdo.org/snomed-ct/.

    torrinomedica.it. 2020. “torrinomedica.it - Dizionario dei Farmaci.” https://www.torrinomedica.it/schede-farmaci.

    Denny Vrandečić and Markus Krötzsch. 2014. “Wikidata: A Free Collaborative Knowledgebase.” Commun. ACM 57 (10): 78–85. https://doi.org/10.1145/2629489.

    10.1145/2629489 :

    T. Elizabeth Workman, Yijun Shao, Guy Divita, and Qing Zeng-Treitler. 2019. “An Efficient Prototype Method to Identify and Correct Misspellings in Clinical Text.” BMC Research Notes 12 (1): 1–5.

    Yonghui Wu, S. Trent Rosenbloom, Joshua C Denny, Randolph A Miller, Subramani Mani, Dario A Giuse, and Hua Xu. 2011. “Detecting Abbreviations in Discharge Summaries Using Machine Learning Methods.” In AMIA Annual Symposium Proceedings, 2011:1541. American Medical Informatics Association.

    Benat Zapirain, Eneko Agirre, Lluis Marquez, and Mihai Surdeanu. 2013. “Selectional Preferences for Semantic Role Classification.” Computational Linguistics 39 (3): 631–63.

    Notes de bas de page

    1 Copyright © 2020 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).

    2 Related approaches have been designed as Semantic Role Labeling tasks (Gildea and Jurafsky 2002; Zapirain et al. 2013), but also frame-based approaches have been proposed, paired to deep syntactic analysis, to extract salient information through a template-filling approach (Lesmo, Mazzei, and Radicioni 2009; Gianfelice et al. 2013).

    3 https://git.savannah.gnu.org/git/aspell.git and https://languagetool.org, respectively.

    4 We used the set of stop words made available by Spacy (https://spacy.io/) for the Italian language.

    Auteurs

    • Enrico Mensa

      Università di Torino, Dipartimento di Informatica – enrico.mensa@unito.it

    • Gian Manuel Marino

      Università di Torino, Dipartimento di Informatica – marino.jnf@gmail.com

    • Davide Colla

      Università di Torino, Dipartimento di Informatica – davide.colla@unito.it

    • Matteo Delsanto

      Università di Torino, Dipartimento di Informatica – matteo.delsanto@unito.it

    • Daniele P. Radicioni

      Università di Torino, Dipartimento di Informatica – daniele.radicioni@unito.it

    Précédent Suivant
    Table des matières

    Le texte seul est utilisable sous licence Licence OpenEdition Books. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.

    Voir plus de livres
    Proceedings of the Second Italian Conference on Computational Linguistics CLiC-it 2015

    Proceedings of the Second Italian Conference on Computational Linguistics CLiC-it 2015

    3-4 December 2015, Trento

    Cristina Bosco, Sara Tonelli et Fabio Massimo Zanzotto (dir.)

    2015

    Proceedings of the Third Italian Conference on Computational Linguistics CLiC-it 2016

    Proceedings of the Third Italian Conference on Computational Linguistics CLiC-it 2016

    5-6 December 2016, Napoli

    Anna Corazza, Simonetta Montemagni et Giovanni Semeraro (dir.)

    2016

    EVALITA. Evaluation of NLP and Speech Tools for Italian

    EVALITA. Evaluation of NLP and Speech Tools for Italian

    Proceedings of the Final Workshop 7 December 2016, Naples

    Pierpaolo Basile, Franco Cutugno, Malvina Nissim et al. (dir.)

    2016

    Proceedings of the Fourth Italian Conference on Computational Linguistics CLiC-it 2017

    Proceedings of the Fourth Italian Conference on Computational Linguistics CLiC-it 2017

    11-12 December 2017, Rome

    Roberto Basili, Malvina Nissim et Giorgio Satta (dir.)

    2017

    Proceedings of the Fifth Italian Conference on Computational Linguistics CLiC-it 2018

    Proceedings of the Fifth Italian Conference on Computational Linguistics CLiC-it 2018

    10-12 December 2018, Torino

    Elena Cabrio, Alessandro Mazzei et Fabio Tamburini (dir.)

    2018

    EVALITA Evaluation of NLP and Speech Tools for Italian

    EVALITA Evaluation of NLP and Speech Tools for Italian

    Proceedings of the Final Workshop 12-13 December 2018, Naples

    Tommaso Caselli, Nicole Novielli, Viviana Patti et al. (dir.)

    2018

    EVALITA Evaluation of NLP and Speech Tools for Italian - December 17th, 2020

    EVALITA Evaluation of NLP and Speech Tools for Italian - December 17th, 2020

    Proceedings of the Seventh Evaluation Campaign of Natural Language Processing and Speech Tools for Italian Final Workshop

    Valerio Basile, Danilo Croce, Maria Maro et al. (dir.)

    2020

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Bologna, Italy, March 1-3, 2021

    Felice Dell'Orletta, Johanna Monti et Fabio Tamburini (dir.)

    2020

    Proceedings of the Eighth Italian Conference on Computational Linguistics CliC-it 2021

    Proceedings of the Eighth Italian Conference on Computational Linguistics CliC-it 2021

    Milan, Italy, 26-28 January, 2022

    Elisabetta Fersini, Marco Passarotti et Viviana Patti (dir.)

    2022

    Voir plus de livres
    1 / 9
    Proceedings of the Second Italian Conference on Computational Linguistics CLiC-it 2015

    Proceedings of the Second Italian Conference on Computational Linguistics CLiC-it 2015

    3-4 December 2015, Trento

    Cristina Bosco, Sara Tonelli et Fabio Massimo Zanzotto (dir.)

    2015

    Proceedings of the Third Italian Conference on Computational Linguistics CLiC-it 2016

    Proceedings of the Third Italian Conference on Computational Linguistics CLiC-it 2016

    5-6 December 2016, Napoli

    Anna Corazza, Simonetta Montemagni et Giovanni Semeraro (dir.)

    2016

    EVALITA. Evaluation of NLP and Speech Tools for Italian

    EVALITA. Evaluation of NLP and Speech Tools for Italian

    Proceedings of the Final Workshop 7 December 2016, Naples

    Pierpaolo Basile, Franco Cutugno, Malvina Nissim et al. (dir.)

    2016

    Proceedings of the Fourth Italian Conference on Computational Linguistics CLiC-it 2017

    Proceedings of the Fourth Italian Conference on Computational Linguistics CLiC-it 2017

    11-12 December 2017, Rome

    Roberto Basili, Malvina Nissim et Giorgio Satta (dir.)

    2017

    Proceedings of the Fifth Italian Conference on Computational Linguistics CLiC-it 2018

    Proceedings of the Fifth Italian Conference on Computational Linguistics CLiC-it 2018

    10-12 December 2018, Torino

    Elena Cabrio, Alessandro Mazzei et Fabio Tamburini (dir.)

    2018

    EVALITA Evaluation of NLP and Speech Tools for Italian

    EVALITA Evaluation of NLP and Speech Tools for Italian

    Proceedings of the Final Workshop 12-13 December 2018, Naples

    Tommaso Caselli, Nicole Novielli, Viviana Patti et al. (dir.)

    2018

    EVALITA Evaluation of NLP and Speech Tools for Italian - December 17th, 2020

    EVALITA Evaluation of NLP and Speech Tools for Italian - December 17th, 2020

    Proceedings of the Seventh Evaluation Campaign of Natural Language Processing and Speech Tools for Italian Final Workshop

    Valerio Basile, Danilo Croce, Maria Maro et al. (dir.)

    2020

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Bologna, Italy, March 1-3, 2021

    Felice Dell'Orletta, Johanna Monti et Fabio Tamburini (dir.)

    2020

    Proceedings of the Eighth Italian Conference on Computational Linguistics CliC-it 2021

    Proceedings of the Eighth Italian Conference on Computational Linguistics CliC-it 2021

    Milan, Italy, 26-28 January, 2022

    Elisabetta Fersini, Marco Passarotti et Viviana Patti (dir.)

    2022

    Voir plus de chapitres

    Conceptual Abstractness: from Nouns to Verbs

    Davide Colla, Enrico Mensa, Aureliano Porporato et al.

    Taming Sense Sparsity: a Common-Sense Approach

    Antonio Lieto, Enrico Mensa et Daniele P. Radicioni

    CONcreTEXT @ EVALITA2020: The Concreteness in Context Task

    Lorenzo Gregori, Maria Montefinese, Daniele P. Radicioni et al.

    Voir plus de chapitres
    1 / 3

    Conceptual Abstractness: from Nouns to Verbs

    Davide Colla, Enrico Mensa, Aureliano Porporato et al.

    Taming Sense Sparsity: a Common-Sense Approach

    Antonio Lieto, Enrico Mensa et Daniele P. Radicioni

    CONcreTEXT @ EVALITA2020: The Concreteness in Context Task

    Lorenzo Gregori, Maria Montefinese, Daniele P. Radicioni et al.

    Accès ouvert

    Accès ouvert

    ePub

    PDF

    PDF du chapitre

    1 Copyright © 2020 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).

    2 Related approaches have been designed as Semantic Role Labeling tasks (Gildea and Jurafsky 2002; Zapirain et al. 2013), but also frame-based approaches have been proposed, paired to deep syntactic analysis, to extract salient information through a template-filling approach (Lesmo, Mazzei, and Radicioni 2009; Gianfelice et al. 2013).

    3 https://git.savannah.gnu.org/git/aspell.git and https://languagetool.org, respectively.

    4 We used the set of stop words made available by Spacy (https://spacy.io/) for the Italian language.

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    X Facebook Email

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Ce livre est cité par

    • Miaschi, Alessio. Alzetta, Chiara. Brunato, Dominique. Dell’Orletta, Felice. Venturi, Giulia. (2023) Testing the Effectiveness of the Diagnostic Probing Paradigm on Italian Treebanks. Information, 14. DOI: 10.3390/info14030144
    • Miaschi, Alessio. Sarti, Gabriele. Brunato, Dominique. Dell’Orletta, Felice. Venturi, Giulia. (2022) Probing Linguistic Knowledge in Italian Neural Language Models across Language Varieties. Italian Journal of Computational Linguistics, 8. DOI: 10.4000/ijcol.965
    • Saga, Satoshi. Shirakawa, Kanta. (2025) Investigating the Effect of Pseudo-Haptics on Perceptions Toward Onomatopoeia Text During Finger-Point Tracing. Multimodal Technologies and Interaction, 9. DOI: 10.3390/mti9100100
    • (2025) Book reviews. Yearbook of Phraseology, 16. DOI: 10.1515/phras-2025-0012

    Ce chapitre est cité par

    • Mensa, Enrico. Martínez Fernández, Paloma. Roller, Roland. Radicioni, Daniele P.. (2023) Editorial: Information extraction for health documents. Frontiers in Artificial Intelligence, 6. DOI: 10.3389/frai.2023.1224529

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Vous pouvez vous connecter à votre bibliothèque à l’adresse suivante : https://freemium.openedition.org/oebooks

    Suggérer l’acquisition à votre bibliothèque

    Si vous avez des questions, vous pouvez nous écrire à access[at]openedition.org

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Vérifiez si votre bibliothèque a déjà acquis ce livre : authentifiez-vous à OpenEdition Freemium for Books.

    Vous pouvez suggérer à votre bibliothèque d’acquérir un ou plusieurs livres publiés sur OpenEdition Books. N’hésitez pas à lui indiquer nos coordonnées : access[at]openedition.org

    Vous pouvez également nous indiquer, à l’aide du formulaire suivant, les coordonnées de votre bibliothèque afin que nous la contactions pour lui suggérer l’achat de ce livre. Les champs suivis de (*) sont obligatoires.

    Veuillez, s’il vous plaît, remplir tous les champs.

    La syntaxe de l’email est incorrecte.

    Référence numérique du chapitre

    Format

    Mensa, E., Marino, G. M., Colla, D., Delsanto, M., & Radicioni, D. P. (2020). A Resource for Detecting Misspellings and Denoising Medical Text Data. In F. Dell’Orletta, J. Monti, & F. Tamburini (éds.), Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020. Torino: Accademia University Press. https://doi.org/10.4000/books.aaccademia.8728
    Mensa, Enrico, Gian Manuel Marino, Davide Colla, Matteo Delsanto, et Daniele P. Radicioni. « A Resource for Detecting Misspellings and Denoising Medical Text Data ». In Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-It 2020, édité par Felice Dell’Orletta, Johanna Monti, et Fabio Tamburini. Torino: Accademia University Press, 2020. doi:10.4000/books.aaccademia.8728.
    Mensa, Enrico, et al. « A Resource for Detecting Misspellings and Denoising Medical Text Data ». Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-It 2020, édité par Felice Dell’Orletta et al., Accademia University Press, 2020, https://doi.org/10.4000/books.aaccademia.8728.

    Référence numérique du livre

    Format

    Dell’Orletta, F., Monti, J., & Tamburini, F. (éds.). (2020). Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020. Torino: Accademia University Press. https://doi.org/10.4000/books.aaccademia.8203
    Dell’Orletta, Felice, Johanna Monti, et Fabio Tamburini, éd. Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-It 2020. Torino: Accademia University Press, 2020. doi:10.4000/books.aaccademia.8203.
    Dell’Orletta, Felice, et al., éditeurs. Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-It 2020. Accademia University Press, 2020, https://doi.org/10.4000/books.aaccademia.8203.
    Compatible avec Zotero Zotero

    1 / 3

    Accademia University Press

    Accademia University Press

    • Plan du site
    • Se connecter

    Suivez-nous

    • Facebook
    • Flux RSS

    URL : http://www.aaccademia.it/

    Email : info@aaccademia.it

    Adresse :

    Accademia University Press

    Via Carlo Alberto 55

    I‐10123

    Torino

    Italia

    OpenEdition
    • Candidater à OpenEdition Books
    • Connaître le programme OpenEdition Freemium
    • Commander des livres
    • S’abonner à la lettre d’OpenEdition
    • CGU d’OpenEdition Books
    • Accessibilité : partiellement conforme
    • Données personnelles
    • Gestion des cookies
    • Système de signalement