The tagged LOB Corpus : user's manual

By S. Johansson, Eric Atwell, R. Garside, Geoffrey Leech. 1986

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Long-short term memory neural networks language modeling for handwriting recognition (2012, methodology): “The database consists of texts from the LOB corpus [6] and is separated into writer disjunct training, validation, and testing sets with a size of 6,161, 920, and 2,781 text lines, respectively.”
Improving Offline Handwritten Text Recognition with Hybrid HMM/ANN Models (2011, methodology): “The sentences have been extracted from the Lancaster-Oslo/Bergen (LOB) text corpus [47].”
A Novel Connectionist System for Unconstrained Handwriting Recognition (2009, methodology): “Both databases were created using texts from the Lancaster-Oslo/Bergen (LOB) corpus [50] as prompts.”
Fast Evaluation of Connectionist Language Models (2009, methodology): “In order to test the proposed technique, experiments with the LOB text corpus [8] have been conducted.”
Unconstrained Online Handwriting Recognition with Recurrent Neural Networks (2007, methodology): “The writers were asked to write forms from the LOB text corpus [8], and the position of their pen was tracked using an infra-red device in the corner of the board.”
Supervised Sequence Labelling with Recurrent Neural Networks (2012, methodology): “The writers were asked to write forms from the LOB text corpus (Johansson et al., 1986), and the position of their pen was tracked using an infra-red device in the corner of the board.”
Neural network language models for off-line handwriting recognition (2014, methodology): “The word language models are trained with three different text corpora: the LOB corpus [21] (excluding those sentences that contain some line from the test set or the validation set of the IAM task), the Brown corpus [22], and the Wellington corpus [23].”
Unconstrained On-line Handwriting Recognition with Recurrent Neural Networks (2007, background): “Connectionist temporal classification (CTC) [3] is an objective function designed for sequence labelling with RNNs. Unlike previous objective functions it does not require pre-segmented training data, or post-processing to transform the network outputs into transcriptions.”
Bidirectional Language Model for Handwriting Recognition (2012, methodology): “Both the forward and the backward N -grams with N = 2, 3, 4 are estimated on the union of the Brown and Wellington corpus [1, 8, 9] as well as the part of the LOB corpus not used in the validation or testing.”
Cross-entropy vs. squared error training: a theoretical and experimental comparison (2013, methodology): “We use a 3-gram LM built upon the LOB, Brown, and Wellington corpora [10, 12, 13] with a vocabulary containing the 50k most frequent words.”
IAM-OnDB - an on-line English sentence database acquired from handwritten text on a whiteboard (2005, background): “In off-line recognition the handwritingof a user is given in terms of a static image, while in the online mode it is a time dependent signal that represents the location of the tip of the pen as a user is writing.”
Is There a Core General Vocabulary? Introducing the "New General Service List". (2015, background): “While LOB and its ‘mirror’ corpus BE06 both consist of 500 samples from 15 different genres of writing, each sample comprising 2,000 words (cf. Johansson et al. 1986; Baker 2009), the BNC samples (>4,000 altogether) are usually longer, ranging between 40 and 50 thousand tokens (cf. Aston and…”