Overview of BioCreative II gene mention recognition

Smith, Larry; Tanabe, Lorraine; Ando, R.; Kuo, Chung‐Hao; Chung, I‐Fang; Hsu, Chun‐Nan; Lin, Yutong; Klinger, Roman; Friedrich, Christoph M.; Ganchev, Kuzman; Torii, Manabu; Liu, Hongfang; Haddow, Barry; Struble, Craig A.; Povinelli, Richard J.; Vlachos, Andreas; Baumgartner, William A.; Hunter, Lawrence; Carpenter, Bob; Tsai, Richard Tzong-Han; Dai, Hong-Jie; Feng, Liang; Chen, Yifei; Sun, Chengjie; Katrenko, Sophia; Adriaans, Pieter; Blaschke, Christian; Torres, Rafael; Neves, Mariana; Nakov, Preslav; Divoli, Anna; López, Manuel J. Maña; Mata, Jacinto; Wilbur, W. John

doi:10.1186/gb-2008-9-s2-s2

Cited by 356 publications

(316 citation statements)

References 16 publications

Supporting

Mentioning

314

Contrasting

Order By: Relevance

“…Conditional random fields (CRFs) [15] achieve state-ofthe-art performance across a broad spectrum of sequence labeling tasks, including the gene mention task in the BioCreativeII competition [13]. Linear-chain CRFs are discriminative probabilistic models over observation sequences x and label sequences y = (y 1 , ..., y |y| ), where |x| = |y|, and each label y i has K different possible discrete values (multi-class).…”

Section: B Conditional Random Fieldmentioning

confidence: 99%

“…We then test SLF using conditional random fields (using CRF++ tool) as a base classifier on the gene-namerecognition (GM) data set (the number of tokens in training/testing/unlabeled sets in Table III) from the 2006 BioCreativeII BioNLP competition [13]. For this GM task, we utilized the following word features: (i) words in a 5-word-window surrounding current, (ii) capitalization features of current and surrounding words, (iii) prefix and suffix (up to length 4) of current and surrounding words, (iv) string patterns of current and surrounding words.…”

Section: Experiments a Datasets And Settingsmentioning

confidence: 99%

“…The BioCreativeII GM [13] data set involves a sequence tagging task which looks for gene/protein names in a bioliterature text corpus. Since almost all the top teams in Here we used only basic word, "caps" features, string patterns and prefix/suffix attributes.…”

Section: Clustered and Boundary Slf: Comparison With Crfmentioning

confidence: 99%

“…We tested our approach on four classic sequence labeling tasks: two CoNLL-2003 [11] shared tasks (German and English Named Entity Recognition (NER)), one English Part-of-Speech tagging [12] and the GM BioCreativeII competition (gene name recognition) [13]. Two state-of-the-art natural language processing systems are used as baseline supervised methods to train SLF: (a) a neural network (NN) model [14] for unified NLP (b) a conditional random field (CRF) [15] which has shown success in many information extraction tasks.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Semi-Supervised Sequence Labeling with Self-Learned Features

Qi¹,

Kuksa

Collobert³

et al. 2009

2009 Ninth IEEE International Conference on Data Mining

View full text Add to dashboard Cite

Abstract-Typical information extraction (IE) systems can be seen as tasks assigning labels to words in a natural language sequence. The performance is restricted by the availability of labeled words. To tackle this issue, we propose a semisupervised approach to improve the sequence labeling procedure in IE through a class of algorithms with self-learned features (SLF). A supervised classifier can be trained with annotated text sequences and used to classify each word in a large set of unannotated sentences. By averaging predicted labels over all cases in the unlabeled corpus, SLF training builds class label distribution patterns for each word (or word attribute) in the dictionary and re-trains the current model iteratively adding these distributions as extra word features. Basic SLF models how likely a word could be assigned to target class types. Several extensions are proposed, such as learning words' class boundary distributions. SLF exhibits robust and scalable behaviour and is easy to tune. We applied this approach on four classical IE tasks: named entity recognition (German and English), part-of-speech tagging (English) and one gene name recognition corpus. Experimental results show effective improvements over the supervised baselines on all tasks. In addition, when compared with the closely related self-training idea, this approach shows favorable advantages.

show abstract

Section: B Conditional Random Fieldmentioning

confidence: 99%

Section: Experiments a Datasets And Settingsmentioning

confidence: 99%

Section: Clustered and Boundary Slf: Comparison With Crfmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Semi-Supervised Sequence Labeling with Self-Learned Features

Qi¹,

Kuksa

Collobert³

et al. 2009

2009 Ninth IEEE International Conference on Data Mining

View full text Add to dashboard Cite

show abstract

“…These linguistic issues are often handled using rules. But, except very few attempts (Califf and Mooney, 1999;Smith et al, 2008), such rules are manually elaborated and texts, which can be processed are necessarily specific and limited. Furthermore, machine learning (ML) based methods such as support vector machines, conditional random fields, etc., (Smith et al, 2008) need many features and their outcomes are not really understandable by a user.…”

Section: Introductionmentioning

confidence: 99%

Combining sequence and itemset mining to discover named entities in biomedical texts: a new type of pattern

Plantevit

Charnois

Kléma

et al. 2009

IJDMMM

View full text Add to dashboard Cite

Biomedical named entity recognition (NER) is a challenging problem. In this paper, we show that mining techniques, such as sequential pattern mining and sequential rule mining, can be useful to tackle this problem but present some limitations. We demonstrate and analyse these limitations and introduce a new kind of pattern called LSR pattern that offers an excellent trade-off between the high precision of sequential rules and the high recall of sequential patterns. We formalise the LSR pattern mining problem first. Then we show how LSR patterns enable us to successfully tackle biomedical NER problems. We report experiments carried out on real datasets that underline the relevance of our proposition.

show abstract

Down‐regulation of miR‐146a‐5p and its potential targets in hepatocellular carcinoma validated by a TCGA‐ and GEO‐based study

Zhang

Liang

et al. 2017

FEBS Open Bio

View full text Add to dashboard Cite

Our previous research has demonstrated that miR‐146a‐5p is down‐regulated in hepatocellular carcinoma (HCC) and might play a tumor‐suppressive role. In this study, we sought to validate the decreased expression with a larger cohort and to explore potential molecular mechanisms. GEO and TCGA databases were used to gather miR‐146a‐5p expression data in HCC, which included 762 HCC and 454 noncancerous liver tissues. A meta‐analysis of the GEO‐based microarrays, TCGA‐based RNA‐seq data, and additional qRT‐PCR data validated the down‐regulation of miR‐146a‐5p in HCC and no publication bias was observed. Integrated genes were generated by overlapping miR‐146a‐5p‐related genes from predicted and formerly reported HCC‐related genes using natural language processing. The overlaps were comprehensively analyzed to discover the potential gene signatures, regulatory pathways, and networks of miR‐146a‐5p in HCC. A total of 251 miR‐146a‐5p potential target genes were predicted by bioinformatics platforms and 104 genes were considered as both HCC‐ and miR‐146a‐5p‐related overlaps. RAC1 was the most connected hub gene for miR‐146a‐5p and four pathways with high enrichment (VEGF signaling pathway, adherens junction, toll‐like receptor signaling pathway, and neurotrophin signaling pathway) were denoted for the overlapped genes. The down‐regulation of miR‐146a‐5p in HCC has been validated with the most complete data possible. The potential gene signatures, regulatory pathways, and networks identified for miR‐146a‐5p in HCC could prove useful for molecular‐targeted diagnostics and therapeutics.

show abstract

Overview of BioCreative II gene mention recognition

Cited by 356 publications

References 16 publications

Semi-Supervised Sequence Labeling with Self-Learned Features

Semi-Supervised Sequence Labeling with Self-Learned Features

Combining sequence and itemset mining to discover named entities in biomedical texts: a new type of pattern

Down‐regulation of miR‐146a‐5p and its potential targets in hepatocellular carcinoma validated by a TCGA‐ and GEO‐based study

Contact Info

Product

Resources

About