High Inter-annotator Agreement Research Articles

Annotation Protocol for Textbook Enrichment with Prerequisite Knowledge Graph

Extracting and formally representing the knowledge embedded in textbooks, such as the concepts explained and the relations between them, can support the provision of advanced knowledge-based services for learning environments and digital libraries. In this paper, we consider a specific type of relation in textbooks referred to as prerequisite relations (PR). PRs represent precedence relations between concepts aimed to provide the reader with the knowledge needed to understand a further concept(s). Their annotation in educational texts produces datasets that can be represented as a graph of concepts connected by PRs. However, building good-quality and reliable datasets of PRs from a textbook is still an open issue, not just for automated annotation methods but even for manual annotation. In turn, the lack of good-quality datasets and well-defined criteria to identify PRs affect the development and validation of automated methods for prerequisite identification. As a contribution to this issue, in this paper, we propose PREAP, a protocol for the annotation of prerequisite relations in textbooks aimed at obtaining reliable annotated data that can be shared, compared, and reused in the research community. PREAP defines a novel textbook-driven annotation method aimed to capture the structure of prerequisites underlying the text. The protocol has been evaluated against baseline methods for manual and automatic annotation. The findings show that PREAP enables the creation of prerequisite knowledge graphs that have higher inter-annotator agreement, accuracy, and alignment with text than the baseline methods. This suggests that the protocol is able to accurately capture the PRs expressed in the text. Furthermore, the findings show that the time required to complete the annotation using PREAP are significantly shorter than with the other manual baseline methods. The paper includes also guidelines for using PREAP in three annotation scenarios, experimentally tested. We also provide example datasets and a user interface that we developed to support prerequisite annotation.

PhenoDEF: a corpus for annotating sentences with information of phenotype definitions in biomedical literature

BackgroundAdverse events induced by drug-drug interactions are a major concern in the United States. Current research is moving toward using electronic health record (EHR) data, including for adverse drug events discovery. One of the first steps in EHR-based studies is to define a phenotype for establishing a cohort of patients. However, phenotype definitions are not readily available for all phenotypes. One of the first steps of developing automated text mining tools is building a corpus. Therefore, this study aimed to develop annotation guidelines and a gold standard corpus to facilitate building future automated approaches for mining phenotype definitions contained in the literature. Furthermore, our aim is to improve the understanding of how these published phenotype definitions are presented in the literature and how we annotate them for future text mining tasks.ResultsTwo annotators manually annotated the corpus on a sentence-level for the presence of evidence for phenotype definitions. Three major categories (inclusion, intermediate, and exclusion) with a total of ten dimensions were proposed characterizing major contextual patterns and cues for presenting phenotype definitions in published literature. The developed annotation guidelines were used to annotate the corpus that contained 3971 sentences: 1923 out of 3971 (48.4%) for the inclusion category, 1851 out of 3971 (46.6%) for the intermediate category, and 2273 out of 3971 (57.2%) for exclusion category. The highest number of annotated sentences was 1449 out of 3971 (36.5%) for the “Biomedical & Procedure” dimension. The lowest number of annotated sentences was 49 out of 3971 (1.2%) for “The use of NLP”. The overall percent inter-annotator agreement was 97.8%. Percent and Kappa statistics also showed high inter-annotator agreement across all dimensions.ConclusionsThe corpus and annotation guidelines can serve as a foundational informatics approach for annotating and mining phenotype definitions in literature, and can be used later for text mining applications.

DravidianCodeMix: sentiment analysis and offensive language identification dataset for Dravidian languages in code-mixed text

This paper describes the development of a multilingual, manually annotated dataset for three under-resourced Dravidian languages generated from social media comments. The dataset was annotated for sentiment analysis and offensive language identification for a total of more than 60,000 YouTube comments. The dataset consists of around 44,000 comments in Tamil-English, around 7000 comments in Kannada-English, and around 20,000 comments in Malayalam-English. The data was manually annotated by volunteer annotators and has a high inter-annotator agreement in Krippendorff’s alpha. The dataset contains all types of code-mixing phenomena since it comprises user-generated content from a multilingual country. We also present baseline experiments to establish benchmarks on the dataset using machine learning and deep learning methods. The dataset is available on Github and Zenodo.

An NLP-based citation reason analysis using CCRO

In recent scientific advances, Artificial Intelligence and Natural Language Processing are the major contributors to classifying documents and extracting information. Classifying citations in different classes have gathered a lot of attention due to the large volume of citations available in different digital libraries. Typical citation classification uses sentiment analysis, where various techniques are applied to citations texts to mainly classify them in “Positive”, “Negative” and “Neutral” sentiments. However, there can be innumerable reasons why an author selects another research for citation. Citations’ Context and Reasons Ontology—CCRO uses a clear scientific method to articulate eight basic reasons for citing by using an iterative process of sentiment analysis, collaborative meanings, and experts' opinions. Using CCRO, this research paper adopts an ontology-based approach to extract citation's reasons and instantiate ontology classes and properties on two different corpora of citation sentences. One corpus of citation sentences is a publicly available dataset, while the other is our own manually curated. The process uses a two-step approach. The first part is an interface to manually annotate each citation text in the selected corpora on CCRO properties. A team of carefully selected annotators has annotated each citation to achieve a high inter-annotator agreement. The second part focuses on the automatic extraction of these reasons. Using Natural Language Processing, Mapping Graph, and Reporting Verb in a citation sentence, citation's reason is extracted and mapped onto a CCRO property. After comparing both manual and automatic mapping, accuracy is calculated. Based on experiments and results, accuracy is calculated for both publicly available and own corpora of citation sentences.

Lexicon Development for COVID-19-related Concepts Using Open-source Word Embedding Sources: An Intrinsic and Extrinsic Evaluation.

BackgroundScientists are developing new computational methods and prediction models to better clinically understand COVID-19 prevalence, treatment efficacy, and patient outcomes. These efforts could be improved by leveraging documented COVID-19–related symptoms, findings, and disorders from clinical text sources in an electronic health record. Word embeddings can identify terms related to these clinical concepts from both the biomedical and nonbiomedical domains, and are being shared with the open-source community at large. However, it’s unclear how useful openly available word embeddings are for developing lexicons for COVID-19–related concepts.ObjectiveGiven an initial lexicon of COVID-19–related terms, this study aims to characterize the returned terms by similarity across various open-source word embeddings and determine common semantic and syntactic patterns between the COVID-19 queried terms and returned terms specific to the word embedding source.MethodsWe compared seven openly available word embedding sources. Using a series of COVID-19–related terms for associated symptoms, findings, and disorders, we conducted an interannotator agreement study to determine how accurately the most similar returned terms could be classified according to semantic types by three annotators. We conducted a qualitative study of COVID-19 queried terms and their returned terms to detect informative patterns for constructing lexicons. We demonstrated the utility of applying such learned synonyms to discharge summaries by reporting the proportion of patients identified by concept among three patient cohorts: pneumonia (n=6410), acute respiratory distress syndrome (n=8647), and COVID-19 (n=2397).ResultsWe observed high pairwise interannotator agreement (Cohen kappa) for symptoms (0.86-0.99), findings (0.93-0.99), and disorders (0.93-0.99). Word embedding sources generated based on characters tend to return more synonyms (mean count of 7.2 synonyms) compared to token-based embedding sources (mean counts range from 2.0 to 3.4). Word embedding sources queried using a qualifier term (eg, dry cough or muscle pain) more often returned qualifiers of the similar semantic type (eg, “dry” returns consistency qualifiers like “wet” and “runny”) compared to a single term (eg, cough or pain) queries. A higher proportion of patients had documented fever (0.61-0.84), cough (0.41-0.55), shortness of breath (0.40-0.59), and hypoxia (0.51-0.56) retrieved than other clinical features. Terms for dry cough returned a higher proportion of patients with COVID-19 (0.07) than the pneumonia (0.05) and acute respiratory distress syndrome (0.03) populations.ConclusionsWord embeddings are valuable technology for learning related terms, including synonyms. When leveraging openly available word embedding sources, choices made for the construction of the word embeddings can significantly influence the words learned.

Annotation and extraction of age and temporally-related events from clinical histories

BackgroundAge and time information stored within the histories of clinical notes can provide valuable insights for assessing a patient’s disease risk, understanding disease progression, and studying therapeutic outcomes. However, details of age and temporally-specified clinical events are not well captured, consistently codified, and readily available to research databases for study.MethodsWe expanded upon existing annotation schemes to capture additional age and temporal information, conducted an annotation study to validate our expanded schema, and developed a prototypical, rule-based Named Entity Recognizer to extract our novel clinical named entities (NE). The annotation study was conducted on 138 discharge summaries from the pre-annotated 2014 ShARe/CLEF eHealth Challenge corpus. In addition to existing NE classes (TIMEX3, SUBJECT_CLASS, DISEASE_DISORDER), our schema proposes 3 additional NEs (AGE, PROCEDURE, OTHER_EVENTS). We also propose new attributes, e.g., “degree_relation” which captures the degree of biological relation for subjects annotated under SUBJECT_CLASS. As a proof of concept, we applied the schema to 49 H&P notes to encode pertinent history information for a lung cancer cohort study.ResultsAn abundance of information was captured under the new OTHER_EVENTS, PROCEDURE and AGE classes, with 23%, 10% and 8% of all annotated NEs belonging to the above classes, respectively. We observed high inter-annotator agreement of >80% for AGE and TIMEX3; the automated NLP system achieved F1 scores of 86% (AGE) and 86% (TIMEX3). Age and temporally-specified mentions within past medical, family, surgical, and social histories were common in our lung cancer data set; annotation is ongoing to support this translational research study.ConclusionsOur annotation schema and NLP system can encode historical events from clinical notes to support clinical and translational research studies.

#MeTooMA: Multi-Aspect Annotations of Tweets Related to the MeToo Movement

In this paper, we present a dataset containing 9,973 tweets related to the MeToo movement that were manually annotated for five different linguistic aspects: relevance, stance, hate speech, sarcasm, and dialogue acts. We present a detailed account of the data collection and annotation processes. The annotations have a very high inter-annotator agreement (0.79 to 0.93 k-alpha) due to the domain expertise of the annotators and clear annotation instructions. We analyze the data in terms of geographical distribution, label correlations, and keywords. Lastly, we present some potential use cases of this dataset. We expect this dataset would be of great interest to psycholinguists, socio-linguists, and computational linguists to study the discursive space of digitally mobilized social movements on sensitive issues like sexual harassment.

Fine-Grained Argument Unit Recognition and Classification

Prior work has commonly defined argument retrieval from heterogeneous document collections as a sentence-level classification task. Consequently, argument retrieval suffers both from low recall and from sentence segmentation errors making it difficult for humans and machines to consume the arguments. In this work, we argue that the task should be performed on a more fine-grained level of sequence labeling. For this, we define the task as Argument Unit Recognition and Classification (AURC). We present a dataset of arguments from heterogeneous sources annotated as spans of tokens within a sentence, as well as with a corresponding stance. We show that and how such difficult argument annotations can be effectively collected through crowdsourcing with high inter-annotator agreement. The new benchmark, AURC-8, contains up to 15% more arguments per topic as compared to annotations on the sentence level. We identify a number of methods targeted at AURC sequence labeling, achieving close to human performance on known domains. Further analysis also reveals that, contrary to previous approaches, our methods are more robust against sentence segmentation errors. We publicly release our code and the AURC-8 dataset.1

Open Broadcast Media Audio from TV: A Dataset of TV Broadcast Audio with Relative Music Loudness Annotations

Open Broadcast Media Audio from TV (OpenBMAT) is an open, annotated dataset for the task of music detection that contains over 27 hours of TV broadcast audio from 4 countries distributed over 1647 one-minute long excerpts. It is designed to encompass several essential features for any music detection dataset and is the first one to include annotations about the loudness of music in relation to other simultaneous non-music sounds. OpenBMAT has been cross-annotated by 3 annotators obtaining high inter-annotator agreement percentages, which allows us to validate the annotation methodology and ensure the annotations reliability. In this work, we first review the current publicly available music detection datasets and state OpenBMAT’s contributions. After that, we detail its building process: the selection of the audio and the annotation methodology. Then, we analyze the produced annotations and validate their reliability. We continue with an experiment to highlight the value of these annotations and investigate the most challenging content in OpenBMAT. Finally, we describe the details about the format in which the dataset is presented and the platform where we have made it available. We believe OpenBMAT will contribute to major advancements of the research on music detection in real-life scenarios.

Computational text analysis within the Humanities: How to combine working practices from the contributing fields?

This position paper is based on a keynote presentation at the COLING 2016 Workshop on Language Technology for Digital Humanities in Osaka, Japan. It departs from observations about working practices in Humanities disciplines following a hermeneutic tradition of text interpretation versus the method-oriented research strategies in Computational Linguistics (CL). The respective praxeological traditions are quite different. Yet more and more researchers are willing to open up towards truly transdisciplinary collaborations, trying to exploit advanced methods from CL within research that ultimately addresses questions from the traditional Humanities disciplines and the Social Sciences. The article identifies two central workflow-related issues for this type of collaborative project in the Digital Humanities (DH) and Computational Social Science: (1) a scheduling dilemma, which affects the point in the course of the project when specifications of the core analysis task are fixed (as early as possible from the computational perspective, but as late as possible from the Humanities perspective) and (2) the subjectivity problem, which concerns the degree of intersubjective stability of the target categories of analysis. CL methodology demands high inter-annotator agreement and theory-independent categories, while the categories in hermeneutic reasoning are often tied to a particular interpretive approach (viz. a theory of literary interpretation) and may bear a non-trivial relation to a reader’s pre-understanding. Building a comprehensive methodological framework that helps overcome these issues requires considerable time and patience. The established computational methodology has to be gradually opened up to more hermeneutically oriented research questions; resources and tools for the relevant categories of analysis have to be constructed. This article does not call into question that well-targeted efforts along this path are worthwhile. Yet, it makes the following additional programmatic point regarding directions for future research: It might be fruitful to explore—in parallel—the potential lying in DH-specific variants of the concept of rapid prototyping from Software Engineering. To get an idea of how computational analysis of some aspect of text might contribute to a hermeneutic research question, a prototypical analysis model is constructed, e.g., from related data collections and analysis categories, using transfer techniques. While the initial quality of analysis may be limited, the idea of rapid probing allows scholars to explore how the analysis fits in an actual workflow on the target text data and it can thus provide early feedback for the process of refining the modeling. If the rapid probing method can indeed be incorporated in a hermeneutic framework to the satisfaction of well-disposed Humanities scholars, a swifter exploration of alternative paths of analysis would become possible. This may generate considerable additional momentum for transdisciplinary integration. It is as yet too early to point to truly Humanities-oriented examples of the proposed rapid probing technique. To nevertheless make the programmatic idea more concrete, the article uses two experimental scenarios to argue how rapid probing might help addressing the scheduling dilemma and the subjectivity problem respectively. The first scenario illustrates the transfer of complex analysis pipelines across corpora; the second one addresses rapid annotation experiments targeting character mentions in literary text.

Comparing the Utility of Different Classification Schemes for Emotive Language Analysis

In this paper we investigated the utility of different classification schemes for emotive language analysis with the aim of providing experimental justification for the choice of scheme for classifying emotions in free text. We compared six schemes: (1) Ekman's six basic emotions, (2) Plutchik's wheel of emotion, (3) Watson and Tellegen's Circumplex theory of affect, (4) the Emotion Annotation Representation Language (EARL), (5) WordNet–Affect, and (6) free text. To measure their utility, we investigated their ease of use by human annotators as well as the performance of supervised machine learning. We assembled a corpus of 500 emotionally charged text documents. The corpus was annotated manually using an online crowdsourcing platform with five independent annotators per document. Assuming that classification schemes with a better balance between completeness and complexity are easier to interpret and use, we expect such schemes to be associated with higher inter–annotator agreement. We used Krippendorff's alpha coefficient to measure inter–annotator agreement according to which the six classification schemes were ranked as follows: (1) six basic emotions (α = 0.483), (2) wheel of emotion (α = 0.410), (3) Circumplex (α = 0.312), EARL (α = 0.286), (5) free text (α = 0.205), and (6) WordNet–Affect (α = 0.202). However, correspondence analysis of annotations across the schemes highlighted that basic emotions are oversimplified representations of complex phenomena and as such likely to lead to invalid interpretations, which are not necessarily reflected by high inter-annotator agreement. To complement the result of the quantitative analysis, we used semi–structured interviews to gain a qualitative insight into how annotators interacted with and interpreted the chosen schemes. The size of the classification scheme was highlighted as a significant factor affecting annotation. In particular, the scheme of six basic emotions was perceived as having insufficient coverage of the emotion space forcing annotators to often resort to inferior alternatives, e.g. using happiness as a surrogate for love. On the opposite end of the spectrum, large schemes such as WordNet–Affect were linked to choice fatigue, which incurred significant cognitive effort in choosing the best annotation. In the second part of the study, we used the annotated corpus to create six training datasets, one for each scheme. The training data were used in cross–validation experiments to evaluate classification performance in relation to different schemes. According to the F-measure, the classification schemes were ranked as follows: (1) six basic emotions (F = 0.410), (2) Circumplex (F = 0.341), (3) wheel of emotion (F = 0.293), (4) EARL (F = 0.254), (5) free text (F = 0.159) and (6) WordNet–Affect (F = 0.158). Not surprisingly, the smallest scheme was ranked the highest in both criteria. Therefore, out of the six schemes studied here, six basic emotions are best suited for emotive language analysis. However, both quantitative and qualitative analysis highlighted its major shortcoming – oversimplification of positive emotions, which are all conflated into happiness. Further investigation is needed into ways of better balancing positive and negative emotions.

Towards Human–Machine Collaboration in Creating an Evaluation Corpus for Adverse Drug Events in Discharge Summaries of Electronic Medical Records

Adverse drug events (ADEs) contribute significantly to morbidity and mortality in the healthcare system. The availability of digitalised hospitals' narrative clinical data offers a potentially rich resource to enhance pharmacovigilance efforts to manage potential safety issues arising from real-world use of drugs. The goal of this paper was to establish a foundation for creating an evaluation corpus by developing a set of annotation guidelines to achieve high inter-annotator agreement (IAA) and to evaluate the performance of basic entity identification tools for drugs, adverse events (AEs) and drug-AE relationships from 100 discharge summaries of a tertiary hospital in Singapore. Two teams of three annotators worked independently on text annotation using Knowtator. Three-way IAA of 86%, 70% and 49% were achieved for drugs, AEs and drug-AE relationships respectively. The performance of the machine algorithm was evaluated against annotations made by at least two annotators, with a recall of 84% and precision of 73% for drugs and a recall of 67% and precision of 53% for AEs. The high recall and precision for drug entity extraction suggests that machine pre-annotation of drugs followed by human annotation of AEs and drug-AE relationships could be a feasible approach in expediting the process of creating a larger evaluation corpus. Non-matches between machine and human annotations were examined to identify ways to further refine the algorithm. When successfully implemented, the identification of ADEs could greatly support pharmacovigilance work in characterising the magnitude and scope of ADEs and prioritising interventions to improve the drug safety.

Annotated Corpus for Citation Context Analysis

In this paper, we present a corpus composed of 85 scientific articles annotated with 2092 citations analyzed using context analysis. We obtained a high Inter-annotator agreement; therefore, we assure reliability and reproducibility of the annotation performed by three coders in an independent way. We applied this corpus to classify citations according to qualitative criteria using a medium granularity categorization scheme enriched by annotated keywords and labels to obtain high granularity. The annotation schema handle three dimensions: PURPOSE: POLARITY: ASPECTS. Citation purpose define functions classification: use, critique, comparison and background with more specific classes stablished using keywords: Based on, Supply; Useful; Contrast; Acknowledge, Corroboration, Debate; Weakness and Hedges. Citation aspects complement the citation characterization: concept, method, data, tool, task, among others. Polarity has three levels: Positive, Negative and Neutral. We developed the schema and annotated the corpus focusing in applications for citation influence assessment, but we suggest that applications as summary generation and information retrieval also could use this annotated corpus because of the organization of the scheme in clearly defined general dimensions.

DREAM: Classification scheme for dialog acts in clinical research query mediation

Clinical data access involves complex but opaque communication between medical researchers and query analysts. Understanding such communication is indispensable for designing intelligent human–machine dialog systems that automate query formulation. This study investigates email communication and proposes a novel scheme for classifying dialog acts in clinical research query mediation. We analyzed 315 email messages exchanged in the communication for 20 data requests obtained from three institutions. The messages were segmented into 1333 utterance units. Through a rigorous process, we developed a classification scheme and applied it for dialog act annotation of the extracted utterances. Evaluation results with high inter-annotator agreement demonstrate the reliability of this scheme. This dataset is used to contribute preliminary understanding of dialog acts distribution and conversation flow in this dialog space.

LIRIS-ACCEDE: A Video Database for Affective Content Analysis

Research in affective computing requires ground truth data for training and benchmarking computational models for machine-based emotion understanding. In this paper, we propose a large video database, namely LIRIS-ACCEDE, for affective content analysis and related applications, including video indexing, summarization or browsing. In contrast to existing datasets with very few video resources and limited accessibility due to copyright constraints, LIRIS-ACCEDE consists of 9,800 good quality video excerpts with a large content diversity. All excerpts are shared under creative commons licenses and can thus be freely distributed without copyright issues. Affective annotations were achieved using crowdsourcing through a pair-wise video comparison protocol, thereby ensuring that annotations are fully consistent, as testified by a high inter-annotator agreement, despite the large diversity of raters' cultural backgrounds. In addition, to enable fair comparison and landmark progresses of future affective computational models, we further provide four experimental protocols and a baseline for prediction of emotions using a large set of both visual and audio features. The dataset (the video clips, annotations, features and protocols) is publicly available at: http://liris-accede.ec-lyon.fr/.

Using annotations on Mechanical Turk to perform supervised polarity classification of Spanish customer comments

One of the major bottlenecks in the development of data-driven AI Systems is the cost of reliable human annotations. The recent advent of several crowdsourcing platforms such as Amazon’s Mechanical Turk, allowing requesters the access to affordable and rapid results of a global workforce, greatly facilitates the creation of massive training data. Most of the available studies on the effectiveness of crowdsourcing report on English data. We use Mechanical Turk annotations to train an Opinion Mining System to classify Spanish consumer comments. We design three different Human Intelligence Task (HIT) strategies and report high inter-annotator agreement between non-experts and expert annotators. We evaluate the advantages/drawbacks of each HIT design and show that, in our case, the use of non-expert annotations is a viable and cost-effective alternative to expert annotations.

NCBI disease corpus: A resource for disease name recognition and concept normalization

Information encoded in natural language in biomedical literature publications is only useful if efficient and reliable ways of accessing and analyzing that information are available. Natural language processing and text mining tools are therefore essential for extracting valuable information, however, the development of powerful, highly effective tools to automatically detect central biomedical concepts such as diseases is conditional on the availability of annotated corpora.This paper presents the disease name and concept annotations of the NCBI disease corpus, a collection of 793 PubMed abstracts fully annotated at the mention and concept level to serve as a research resource for the biomedical natural language processing community. Each PubMed abstract was manually annotated by two annotators with disease mentions and their corresponding concepts in Medical Subject Headings (MeSH®) or Online Mendelian Inheritance in Man (OMIM®). Manual curation was performed using PubTator, which allowed the use of pre-annotations as a pre-step to manual annotations. Fourteen annotators were randomly paired and differing annotations were discussed for reaching a consensus in two annotation phases. In this setting, a high inter-annotator agreement was observed. Finally, all results were checked against annotations of the rest of the corpus to assure corpus-wide consistency.The public release of the NCBI disease corpus contains 6892 disease mentions, which are mapped to 790 unique disease concepts. Of these, 88% link to a MeSH identifier, while the rest contain an OMIM identifier. We were able to link 91% of the mentions to a single disease concept, while the rest are described as a combination of concepts. In order to help researchers use the corpus to design and test disease identification methods, we have prepared the corpus as training, testing and development sets. To demonstrate its utility, we conducted a benchmarking experiment where we compared three different knowledge-based disease normalization methods with a best performance in F-measure of 63.7%. These results show that the NCBI disease corpus has the potential to significantly improve the state-of-the-art in disease name recognition and normalization research, by providing a high-quality gold standard thus enabling the development of machine-learning based approaches for such tasks.The NCBI disease corpus, guidelines and other associated resources are available at: http://www.ncbi.nlm.nih.gov/CBBresearch/Dogan/DISEASE/.

CoMAGC: a corpus with multi-faceted annotations of gene-cancer relations

BackgroundIn order to access the large amount of information in biomedical literature about genes implicated in various cancers both efficiently and accurately, the aid of text mining (TM) systems is invaluable. Current TM systems do target either gene-cancer relations or biological processes involving genes and cancers, but the former type produces information not comprehensive enough to explain how a gene affects a cancer, and the latter does not provide a concise summary of gene-cancer relations.ResultsIn this paper, we present a corpus for the development of TM systems that are specifically targeting gene-cancer relations but are still able to capture complex information in biomedical sentences. We describe CoMAGC, a corpus with multi-faceted annotations of gene-cancer relations. In CoMAGC, a piece of annotation is composed of four semantically orthogonal concepts that together express 1) how a gene changes, 2) how a cancer changes and 3) the causality between the gene and the cancer. The multi-faceted annotations are shown to have high inter-annotator agreement. In addition, we show that the annotations in CoMAGC allow us to infer the prospective roles of genes in cancers and to classify the genes into three classes according to the inferred roles. We encode the mapping between multi-faceted annotations and gene classes into 10 inference rules. The inference rules produce results with high accuracy as measured against human annotations. CoMAGC consists of 821 sentences on prostate, breast and ovarian cancers. Currently, we deal with changes in gene expression levels among other types of gene changes. The corpus is available at http://biopathway.org/CoMAGCunder the terms of the Creative Commons Attribution License ( http://creativecommons.org/licenses/by/2.0).ConclusionsThe corpus will be an important resource for the development of advanced TM systems on gene-cancer relations.

A corpus of full-text journal articles is a robust evaluation tool for revealing differences in performance of biomedical natural language processing tools

BackgroundWe introduce the linguistic annotation of a corpus of 97 full-text biomedical publications, known as the Colorado Richly Annotated Full Text (CRAFT) corpus. We further assess the performance of existing tools for performing sentence splitting, tokenization, syntactic parsing, and named entity recognition on this corpus.ResultsMany biomedical natural language processing systems demonstrated large differences between their previously published results and their performance on the CRAFT corpus when tested with the publicly available models or rule sets. Trainable systems differed widely with respect to their ability to build high-performing models based on this data.ConclusionsThe finding that some systems were able to train high-performing models based on this corpus is additional evidence, beyond high inter-annotator agreement, that the quality of the CRAFT corpus is high. The overall poor performance of various systems indicates that considerable work needs to be done to enable natural language processing systems to work well when the input is full-text journal articles. The CRAFT corpus provides a valuable resource to the biomedical natural language processing community for evaluation and training of new models for biomedical full text publications.

Mining personal experiences and opinions from Web documents

This paper proposes a new UGC-oriented language technology application, which we call experience mining. Experience mining aims at automatically collecting instances of personal experiences as well as opinions from vast amounts of user generated content (UGC) such as weblog and forum posts and storing them in an experience database with semantically rich indices. After discussing the technical issues relating to this new task, we focus on the central problem of factuality analysis, formulate a task definition, and propose a machine learning-based solution. Our empirical evaluation indicates that our factuality analysis defintion is sufficiently well-defined to achieve a high inter-annotator agreement and our Factorial CRF-based model considerably outperforms the baseline. We also present an application system, which currently stores over 50M experience instances extracted from 150M Japanese blog posts with semantic indices and serves an experience search engine for unrestricted users and report on our empirical evaluation of the system's accuracy.