Beyond opinion classification: Extracting facts, opinions and experiences from health forums.

INTRODUCTION: Surveys indicate that patients, particularly those suffering from chronic conditions, strongly benefit from the information found in social networks and online forums. One challenge in accessing online health information is to differentiate between factual and more subjective information. In this work, we evaluate the feasibility of exploiting lexical, syntactic, semantic, network-based and emotional properties of texts to automatically classify patient-generated contents into three types: "experiences", "facts" and "opinions", using machine learning algorithms. In this context, our goal is to develop automatic methods that will make online health information more easily accessible and useful for patients, professionals and researchers. MATERIAL AND METHODS: We work with a set of 3000 posts to online health forums in breast cancer, morbus crohn and different allergies. Each sentence in a post is manually labeled as "experience", "fact" or "opinion". Using this data, we train a support vector machine algorithm to perform classification. The results are evaluated in a 10-fold cross validation procedure. RESULTS: Overall, we find that it is possible to predict the type of information contained in a forum post with a very high accuracy (over 80 percent) using simple text representations such as word embeddings and bags of words. We also analyze more complex features such as those based on the network properties, the polarity of words and the verbal tense of the sentences and show that, when combined with the previous ones, they can boost the results.

Feature engineering for sentiment analysis in e-health forums.

INTRODUCTION: Exploiting information in health-related social media services is of great interest for patients, researchers and medical companies. The challenge is, however, to provide easy, quick and relevant access to the vast amount of information that is available. One step towards facilitating information access to online health data is opinion mining. Even though the classification of patient opinions into positive and negative has been previously tackled, most works make use of machine learning methods and bags of words. Our first contribution is an extensive evaluation of different features, including lexical, syntactic, semantic, network-based, sentiment-based and word embeddings features to represent patient-authored texts for polarity classification. The second contribution of this work is the study of polar facts (i.e. objective information with polar connotations). Traditionally, the presence of polar facts has been neglected and research in polarity classification has been bounded to opinionated texts. We demonstrate the existence and importance of polar facts for the polarity classification of health information. MATERIAL AND METHODS: We annotate a set of more than 3500 posts to online health forums of breast cancer, crohn and different allergies, respectively. Each sentence in a post is manually labeled as "experience", "fact" or "opinion", and as "positive", "negative" and "neutral". Using this data, we train different machine learning algorithms and compare traditional bags of words representations with word embeddings in combination with lexical, syntactic, semantic, network-based and emotional properties of texts to automatically classify patient-authored contents into positive, negative and neutral. Beside, we experiment with a combination of textual and semantic representations by generating concept embeddings using the UMLS Metathesaurus. RESULTS: We reach two main results: first, we find that it is possible to predict polarity of patient-authored contents with a very high accuracy (≈ 70 percent) using word embeddings, and that this considerably outperforms more traditional representations like bags of words; and second, when dealing with medical information, negative and positive facts (i.e. objective information) are nearly as frequent as negative and positive opinions and experiences (i.e. subjective information), and their importance for polarity classification is crucial.

Feature engineering for MEDLINE citation categorization with MeSH.

BACKGROUND: Research in biomedical text categorization has mostly used the bag-of-words representation. Other more sophisticated representations of text based on syntactic, semantic and argumentative properties have been less studied. In this paper, we evaluate the impact of different text representations of biomedical texts as features for reproducing the MeSH annotations of some of the most frequent MeSH headings. In addition to unigrams and bigrams, these features include noun phrases, citation meta-data, citation structure, and semantic annotation of the citations. RESULTS: Traditional features like unigrams and bigrams exhibit strong performance compared to other feature sets. Little or no improvement is obtained when using meta-data or citation structure. Noun phrases are too sparse and thus have lower performance compared to more traditional features. Conceptual annotation of the texts by MetaMap shows similar performance compared to unigrams, but adding concepts from the UMLS taxonomy does not improve the performance of using only mapped concepts. The combination of all the features performs largely better than any individual feature set considered. In addition, this combination improves the performance of a state-of-the-art MeSH indexer. Concerning the machine learning algorithms, we find that those that are more resilient to class imbalance largely obtain better performance. CONCLUSIONS: We conclude that even though traditional features such as unigrams and bigrams have strong performance compared to other features, it is possible to combine them to effectively improve the performance of the bag-of-words representation. We have also found that the combination of the learning algorithm and feature sets has an influence in the overall performance of the system. Moreover, using learning algorithms resilient to class imbalance largely improves performance. However, when using a large set of features, consideration needs to be taken with algorithms due to the risk of over-fitting. Specific combinations of learning algorithms and features for individual MeSH headings could further increase the performance of an indexing system.

Evaluating the use of different positional strategies for sentence selection in biomedical literature summarization.

BACKGROUND: The position of a sentence in a document has been traditionally considered an indicator of the relevance of the sentence, and therefore it is frequently used by automatic summarization systems as an attribute for sentence selection. Sentences close to the beginning of the document are supposed to deal with the main topic and thus are selected for the summary. This criterion has shown to be very effective when summarizing some types of documents, such as news items. However, this property is not likely to be found in other types of documents, such as scientific articles, where other positional criteria may be preferred. The purpose of the present work is to study the utility of different positional strategies for biomedical literature summarization. RESULTS: We have evaluated three different positional strategies: (1) awarding the sentences at the beginning of the document, (2) preferring those at the beginning and end of the document, and (3) weighting the sentences according to the section in which they appear. To this end, we have implemented two summarizers, one based on semantic graphs and the other based on concept frequencies, and evaluated the summaries they produce when combined with each of the positional strategies above using ROUGE metrics. Our results indicate that it is possible to improve the quality of the summaries by weighting the sentences according to the section in which they appear (≈17% improvement in ROUGE-2 for the graph-based summarizer and ≈20% for the frequency-based summarizer), and that the sections containing the more salient information are the Methods and Material and the Discussion and Results ones. CONCLUSIONS: It has been found that the use of traditional positional criteria that award sentences at the beginning and/or the end of the document are not helpful when summarizing scientific literature. In contrast, a more appropriate strategy is that which weights sentences according to the section in which they appear.