text
stringlengths
49
577
label
stringclasses
7 values
metadata
list
This paper proposes an [[ approach ]] to << full parsing >> suitable for Information Extraction from texts .
USED-FOR
[ 4, 4, 6, 7 ]
This paper proposes an approach to [[ full parsing ]] suitable for << Information Extraction >> from texts .
USED-FOR
[ 6, 7, 10, 11 ]
[[ It ]] was implemented in the << IE module >> of FACILE , a EU project for multilingual text classification and IE .
USED-FOR
[ 0, 0, 5, 6 ]
It was implemented in the [[ IE module ]] of << FACILE , a EU project for multilingual text classification and IE >> .
PART-OF
[ 5, 6, 8, 18 ]
It then presents an implemented << graphic interpretation system >> that takes into account a variety of [[ communicative signals ]] , and an evaluation study showing that evidence obtained from shallow processing of the graphic 's caption has a significant impact on the system 's success .
USED-FOR
[ 15, 16, 5, 7 ]
It then presents an implemented graphic interpretation system that takes into account a variety of communicative signals , and an evaluation study showing that evidence obtained from [[ shallow processing ]] of the graphic 's caption has a significant impact on the << system >> 's success .
USED-FOR
[ 27, 28, 40, 40 ]
It then presents an implemented graphic interpretation system that takes into account a variety of communicative signals , and an evaluation study showing that evidence obtained from << shallow processing >> of the [[ graphic 's caption ]] has a significant impact on the system 's success .
USED-FOR
[ 31, 33, 27, 28 ]
[[ Graphical models ]] such as Bayesian Networks -LRB- BNs -RRB- are being increasingly applied to various << computer vision problems >> .
USED-FOR
[ 0, 1, 15, 17 ]
<< Graphical models >> such as [[ Bayesian Networks -LRB- BNs -RRB- ]] are being increasingly applied to various computer vision problems .
HYPONYM-OF
[ 4, 8, 0, 1 ]
One bottleneck in using BN is that learning the << BN model parameters >> often requires a large amount of reliable and [[ representative training data ]] , which proves to be difficult to acquire for many computer vision tasks .
USED-FOR
[ 20, 22, 9, 11 ]
One bottleneck in using BN is that learning the BN model parameters often requires a large amount of reliable and [[ representative training data ]] , which proves to be difficult to acquire for many << computer vision tasks >> .
USED-FOR
[ 20, 22, 33, 35 ]
On the other hand , there is often available [[ qualitative prior knowledge ]] about the << model >> .
FEATURE-OF
[ 9, 11, 14, 14 ]
Such << knowledge >> comes either from [[ domain experts ]] based on their experience or from various physical or geometric constraints that govern the objects we try to model .
USED-FOR
[ 5, 6, 1, 1 ]
Such knowledge comes either from [[ domain experts ]] based on their experience or from various << physical or geometric constraints >> that govern the objects we try to model .
CONJUNCTION
[ 5, 6, 14, 17 ]
Such << knowledge >> comes either from domain experts based on their experience or from various [[ physical or geometric constraints ]] that govern the objects we try to model .
USED-FOR
[ 14, 17, 1, 1 ]
Unlike the [[ quantitative prior ]] , the << qualitative prior >> is often ignored due to the difficulty of incorporating them into the model learning process .
COMPARE
[ 2, 3, 6, 7 ]
Unlike the quantitative prior , the qualitative prior is often ignored due to the difficulty of incorporating [[ them ]] into the << model learning process >> .
PART-OF
[ 17, 17, 20, 22 ]
In this paper , we introduce a closed-form solution to systematically combine the [[ limited training data ]] with some generic << qualitative knowledge >> for BN parameter learning .
CONJUNCTION
[ 13, 15, 19, 20 ]
In this paper , we introduce a closed-form solution to systematically combine the [[ limited training data ]] with some generic qualitative knowledge for << BN parameter learning >> .
USED-FOR
[ 13, 15, 22, 24 ]
In this paper , we introduce a closed-form solution to systematically combine the limited training data with some generic [[ qualitative knowledge ]] for << BN parameter learning >> .
USED-FOR
[ 19, 20, 22, 24 ]
In this paper , we introduce a << closed-form solution >> to systematically combine the limited training data with some generic qualitative knowledge for [[ BN parameter learning ]] .
USED-FOR
[ 22, 24, 7, 8 ]
To validate our method , we compare [[ it ]] with the << Maximum Likelihood -LRB- ML -RRB- estimation method >> under sparse data and with the Expectation Maximization -LRB- EM -RRB- algorithm under incomplete data respectively .
COMPARE
[ 7, 7, 10, 16 ]
To validate our method , we compare [[ it ]] with the Maximum Likelihood -LRB- ML -RRB- estimation method under sparse data and with the << Expectation Maximization -LRB- EM -RRB- algorithm >> under incomplete data respectively .
COMPARE
[ 7, 7, 23, 28 ]
To validate our method , we compare << it >> with the Maximum Likelihood -LRB- ML -RRB- estimation method under [[ sparse data ]] and with the Expectation Maximization -LRB- EM -RRB- algorithm under incomplete data respectively .
USED-FOR
[ 18, 19, 7, 7 ]
To validate our method , we compare it with the << Maximum Likelihood -LRB- ML -RRB- estimation method >> under [[ sparse data ]] and with the Expectation Maximization -LRB- EM -RRB- algorithm under incomplete data respectively .
USED-FOR
[ 18, 19, 10, 16 ]
To validate our method , we compare << it >> with the Maximum Likelihood -LRB- ML -RRB- estimation method under sparse data and with the Expectation Maximization -LRB- EM -RRB- algorithm under [[ incomplete data ]] respectively .
USED-FOR
[ 30, 31, 7, 7 ]
To validate our method , we compare it with the Maximum Likelihood -LRB- ML -RRB- estimation method under sparse data and with the << Expectation Maximization -LRB- EM -RRB- algorithm >> under [[ incomplete data ]] respectively .
USED-FOR
[ 30, 31, 23, 28 ]
To further demonstrate its applications for << computer vision >> , we apply [[ it ]] to learn a BN model for facial Action Unit -LRB- AU -RRB- recognition from real image data .
USED-FOR
[ 11, 11, 6, 7 ]
To further demonstrate its applications for computer vision , we apply [[ it ]] to learn a << BN model >> for facial Action Unit -LRB- AU -RRB- recognition from real image data .
USED-FOR
[ 11, 11, 15, 16 ]
To further demonstrate its applications for computer vision , we apply it to learn a [[ BN model ]] for << facial Action Unit -LRB- AU -RRB- recognition >> from real image data .
USED-FOR
[ 15, 16, 18, 24 ]
To further demonstrate its applications for computer vision , we apply it to learn a BN model for << facial Action Unit -LRB- AU -RRB- recognition >> from [[ real image data ]] .
USED-FOR
[ 26, 28, 18, 24 ]
The experimental results show that with simple and [[ generic qualitative constraints ]] and using only a small amount of << training data >> , our method can robustly and accurately estimate the BN model parameters .
CONJUNCTION
[ 8, 10, 18, 19 ]
The experimental results show that with simple and [[ generic qualitative constraints ]] and using only a small amount of training data , our << method >> can robustly and accurately estimate the BN model parameters .
USED-FOR
[ 8, 10, 22, 22 ]
The experimental results show that with simple and generic qualitative constraints and using only a small amount of [[ training data ]] , our << method >> can robustly and accurately estimate the BN model parameters .
USED-FOR
[ 18, 19, 22, 22 ]
The experimental results show that with simple and generic qualitative constraints and using only a small amount of training data , our [[ method ]] can robustly and accurately estimate the << BN model parameters >> .
USED-FOR
[ 22, 22, 29, 31 ]
In this paper we introduce a [[ modal language LT ]] for imposing << constraints on trees >> , and an extension LT -LRB- LF -RRB- for imposing constraints on trees decorated with feature structures .
USED-FOR
[ 6, 8, 11, 13 ]
In this paper we introduce a modal language LT for imposing constraints on trees , and an [[ extension LT -LRB- LF -RRB- ]] for imposing << constraints on trees decorated with feature structures >> .
USED-FOR
[ 17, 21, 24, 30 ]
The motivation for introducing these [[ languages ]] is to provide tools for formalising << grammatical frameworks >> perspicuously , and the paper illustrates this by showing how the leading ideas of GPSG can be captured in LT -LRB- LF -RRB- .
USED-FOR
[ 5, 5, 12, 13 ]
The motivation for introducing these languages is to provide tools for formalising grammatical frameworks perspicuously , and the paper illustrates this by showing how the leading ideas of [[ GPSG ]] can be captured in << LT -LRB- LF -RRB- >> .
USED-FOR
[ 28, 28, 33, 36 ]
Previous research has demonstrated the utility of [[ clustering ]] in << inducing semantic verb classes >> from undisambiguated corpus data .
USED-FOR
[ 7, 7, 9, 12 ]
Previous research has demonstrated the utility of << clustering >> in inducing semantic verb classes from [[ undisambiguated corpus data ]] .
USED-FOR
[ 14, 16, 7, 7 ]
We describe a new << approach >> which involves [[ clustering subcategorization frame -LRB- SCF -RRB- distributions ]] using the Information Bottleneck and nearest neighbour methods .
PART-OF
[ 7, 13, 4, 4 ]
We describe a new approach which involves << clustering subcategorization frame -LRB- SCF -RRB- distributions >> using the [[ Information Bottleneck and nearest neighbour methods ]] .
USED-FOR
[ 16, 21, 7, 13 ]
A novel [[ evaluation scheme ]] is proposed which accounts for the effect of << polysemy >> on the clusters , offering us a good insight into the potential and limitations of semantically classifying undisambiguated SCF data .
USED-FOR
[ 2, 3, 12, 12 ]
A novel [[ evaluation scheme ]] is proposed which accounts for the effect of polysemy on the clusters , offering us a good insight into the potential and limitations of << semantically classifying undisambiguated SCF data >> .
EVALUATE-FOR
[ 2, 3, 28, 32 ]
A novel evaluation scheme is proposed which accounts for the effect of [[ polysemy ]] on the << clusters >> , offering us a good insight into the potential and limitations of semantically classifying undisambiguated SCF data .
FEATURE-OF
[ 12, 12, 15, 15 ]
Due to the capacity of [[ pan-tilt-zoom -LRB- PTZ -RRB- cameras ]] to simultaneously cover a << panoramic area >> and maintain high resolution imagery , researches in automated surveillance systems with multiple PTZ cameras have become increasingly important .
USED-FOR
[ 5, 9, 14, 15 ]
Due to the capacity of [[ pan-tilt-zoom -LRB- PTZ -RRB- cameras ]] to simultaneously cover a panoramic area and maintain << high resolution imagery >> , researches in automated surveillance systems with multiple PTZ cameras have become increasingly important .
USED-FOR
[ 5, 9, 18, 20 ]
Due to the capacity of pan-tilt-zoom -LRB- PTZ -RRB- cameras to simultaneously cover a panoramic area and maintain high resolution imagery , researches in << automated surveillance systems >> with multiple [[ PTZ cameras ]] have become increasingly important .
FEATURE-OF
[ 29, 30, 24, 26 ]
Most existing [[ algorithms ]] require the prior knowledge of intrinsic parameters of the PTZ camera to infer the << relative positioning >> and orientation among multiple PTZ cameras .
USED-FOR
[ 2, 2, 17, 18 ]
Most existing [[ algorithms ]] require the prior knowledge of intrinsic parameters of the PTZ camera to infer the relative positioning and << orientation >> among multiple PTZ cameras .
USED-FOR
[ 2, 2, 20, 20 ]
Most existing << algorithms >> require the [[ prior knowledge of intrinsic parameters of the PTZ camera ]] to infer the relative positioning and orientation among multiple PTZ cameras .
USED-FOR
[ 5, 13, 2, 2 ]
Most existing algorithms require the prior knowledge of intrinsic parameters of the PTZ camera to infer the [[ relative positioning ]] and << orientation >> among multiple PTZ cameras .
CONJUNCTION
[ 17, 18, 20, 20 ]
Most existing algorithms require the prior knowledge of intrinsic parameters of the PTZ camera to infer the [[ relative positioning ]] and orientation among multiple << PTZ cameras >> .
FEATURE-OF
[ 17, 18, 23, 24 ]
Most existing algorithms require the prior knowledge of intrinsic parameters of the PTZ camera to infer the relative positioning and [[ orientation ]] among multiple << PTZ cameras >> .
FEATURE-OF
[ 20, 20, 23, 24 ]
To overcome this limitation , we propose a novel [[ mapping algorithm ]] that derives the << relative positioning >> and orientation between two PTZ cameras based on a unified polynomial model .
USED-FOR
[ 9, 10, 14, 15 ]
To overcome this limitation , we propose a novel [[ mapping algorithm ]] that derives the relative positioning and << orientation >> between two PTZ cameras based on a unified polynomial model .
USED-FOR
[ 9, 10, 17, 17 ]
To overcome this limitation , we propose a novel mapping algorithm that derives the [[ relative positioning ]] and << orientation >> between two PTZ cameras based on a unified polynomial model .
CONJUNCTION
[ 14, 15, 17, 17 ]
To overcome this limitation , we propose a novel mapping algorithm that derives the [[ relative positioning ]] and orientation between two << PTZ cameras >> based on a unified polynomial model .
FEATURE-OF
[ 14, 15, 20, 21 ]
To overcome this limitation , we propose a novel mapping algorithm that derives the relative positioning and [[ orientation ]] between two << PTZ cameras >> based on a unified polynomial model .
FEATURE-OF
[ 17, 17, 20, 21 ]
To overcome this limitation , we propose a novel << mapping algorithm >> that derives the relative positioning and orientation between two PTZ cameras based on a [[ unified polynomial model ]] .
USED-FOR
[ 25, 27, 9, 10 ]
Experimental results demonstrate that our proposed << algorithm >> presents substantially reduced [[ computational complexity ]] and improved flexibility at the cost of slightly decreased pixel accuracy , as compared with the work of Chen and Wang .
EVALUATE-FOR
[ 10, 11, 6, 6 ]
Experimental results demonstrate that our proposed << algorithm >> presents substantially reduced computational complexity and improved [[ flexibility ]] at the cost of slightly decreased pixel accuracy , as compared with the work of Chen and Wang .
EVALUATE-FOR
[ 14, 14, 6, 6 ]
Experimental results demonstrate that our proposed << algorithm >> presents substantially reduced computational complexity and improved flexibility at the cost of slightly decreased [[ pixel accuracy ]] , as compared with the work of Chen and Wang .
EVALUATE-FOR
[ 21, 22, 6, 6 ]
This slightly decreased << pixel accuracy >> can be compensated by [[ consistent labeling approaches ]] without added cost for the application of automated surveillance systems along with changing configurations and a larger number of PTZ cameras .
USED-FOR
[ 9, 11, 3, 4 ]
This paper presents a new [[ two-pass algorithm ]] for << Extra Large -LRB- more than 1M words -RRB- Vocabulary COntinuous Speech recognition >> based on the Information Retrieval -LRB- ELVIRCOS -RRB- .
USED-FOR
[ 5, 6, 8, 19 ]
This paper presents a new << two-pass algorithm >> for Extra Large -LRB- more than 1M words -RRB- Vocabulary COntinuous Speech recognition based on the [[ Information Retrieval -LRB- ELVIRCOS -RRB- ]] .
USED-FOR
[ 23, 27, 5, 6 ]
The principle of this approach is to decompose a recognition process into two << passes >> where the [[ first pass ]] builds the words subset for the second pass recognition by using information retrieval procedure .
HYPONYM-OF
[ 16, 17, 13, 13 ]
The principle of this approach is to decompose a recognition process into two << passes >> where the first pass builds the words subset for the [[ second pass recognition ]] by using information retrieval procedure .
HYPONYM-OF
[ 24, 26, 13, 13 ]
The principle of this approach is to decompose a recognition process into two passes where the first pass builds the words subset for the << second pass recognition >> by using [[ information retrieval procedure ]] .
USED-FOR
[ 29, 31, 24, 26 ]
[[ Word graph composition ]] for << continuous speech >> is presented .
USED-FOR
[ 0, 2, 4, 5 ]
With this [[ approach ]] a high performances for << large vocabulary speech recognition >> can be obtained .
USED-FOR
[ 2, 2, 7, 10 ]
First , images are partitioned into regions using << one-class classification >> and [[ patch-based clustering algorithms ]] where one-class classifiers model the regions with relatively uniform color and texture properties , and clustering of patches aims to detect structures in the remaining regions .
CONJUNCTION
[ 11, 13, 8, 9 ]
First , images are partitioned into regions using one-class classification and patch-based clustering algorithms where << one-class classifiers >> model the regions with relatively [[ uniform color and texture properties ]] , and clustering of patches aims to detect structures in the remaining regions .
USED-FOR
[ 22, 26, 15, 16 ]
Next , the resulting regions are clustered to obtain a codebook of region types , and two [[ models ]] are constructed for << scene representation >> : a '' bag of individual regions '' representation where each region is regarded separately , and a '' bag of region pairs '' representation where regions with particular...
USED-FOR
[ 17, 17, 21, 22 ]
Given these representations , << scene classification >> is done using [[ Bayesian classifiers ]] .
USED-FOR
[ 9, 10, 4, 5 ]
Experiments on the [[ LabelMe data set ]] showed that the proposed << models >> significantly out-perform a baseline global feature-based approach .
EVALUATE-FOR
[ 3, 5, 10, 10 ]
Experiments on the [[ LabelMe data set ]] showed that the proposed models significantly out-perform a << baseline global feature-based approach >> .
EVALUATE-FOR
[ 3, 5, 14, 17 ]
Experiments on the LabelMe data set showed that the proposed [[ models ]] significantly out-perform a << baseline global feature-based approach >> .
COMPARE
[ 10, 10, 14, 17 ]
The [[ model ]] is designed for use in << error correction >> , with a focus on post-processing the output of black-box OCR systems in order to make it more useful for NLP tasks .
USED-FOR
[ 1, 1, 7, 8 ]
The [[ model ]] is designed for use in error correction , with a focus on << post-processing >> the output of black-box OCR systems in order to make it more useful for NLP tasks .
USED-FOR
[ 1, 1, 14, 14 ]
The model is designed for use in << error correction >> , with a focus on [[ post-processing ]] the output of black-box OCR systems in order to make it more useful for NLP tasks .
PART-OF
[ 14, 14, 7, 8 ]
The model is designed for use in error correction , with a focus on post-processing the output of black-box OCR systems in order to make [[ it ]] more useful for << NLP tasks >> .
USED-FOR
[ 25, 25, 29, 30 ]
We present an implementation of the << model >> based on [[ finite-state models ]] , demonstrate the model 's ability to significantly reduce character and word error rate , and provide evaluation results involving automatic extraction of translation lexicons from printed text .
USED-FOR
[ 9, 10, 6, 6 ]
We present an implementation of the model based on finite-state models , demonstrate the << model >> 's ability to significantly reduce [[ character and word error rate ]] , and provide evaluation results involving automatic extraction of translation lexicons from printed text .
EVALUATE-FOR
[ 20, 24, 14, 14 ]
We present an implementation of the model based on finite-state models , demonstrate the << model >> 's ability to significantly reduce character and word error rate , and provide evaluation results involving [[ automatic extraction of translation lexicons ]] from printed text .
EVALUATE-FOR
[ 31, 35, 14, 14 ]
We present an implementation of the model based on finite-state models , demonstrate the model 's ability to significantly reduce character and word error rate , and provide evaluation results involving << automatic extraction of translation lexicons >> from [[ printed text ]] .
USED-FOR
[ 37, 38, 31, 35 ]
We present a [[ framework ]] for << word alignment >> based on log-linear models .
USED-FOR
[ 3, 3, 5, 6 ]
We present a << framework >> for word alignment based on [[ log-linear models ]] .
USED-FOR
[ 9, 10, 3, 3 ]
All [[ knowledge sources ]] are treated as << feature functions >> , which depend on the source langauge sentence , the target language sentence and possible additional variables .
USED-FOR
[ 1, 2, 6, 7 ]
[[ Log-linear models ]] allow << statistical alignment models >> to be easily extended by incorporating syntactic information .
USED-FOR
[ 0, 1, 3, 5 ]
<< Log-linear models >> allow statistical alignment models to be easily extended by incorporating [[ syntactic information ]] .
USED-FOR
[ 12, 13, 0, 1 ]
In this paper , we use [[ IBM Model 3 alignment probabilities ]] , << POS correspondence >> , and bilingual dictionary coverage as features .
CONJUNCTION
[ 6, 10, 12, 13 ]
In this paper , we use [[ IBM Model 3 alignment probabilities ]] , POS correspondence , and bilingual dictionary coverage as << features >> .
USED-FOR
[ 6, 10, 20, 20 ]
In this paper , we use IBM Model 3 alignment probabilities , [[ POS correspondence ]] , and << bilingual dictionary coverage >> as features .
CONJUNCTION
[ 12, 13, 16, 18 ]
In this paper , we use IBM Model 3 alignment probabilities , [[ POS correspondence ]] , and bilingual dictionary coverage as << features >> .
USED-FOR
[ 12, 13, 20, 20 ]
In this paper , we use IBM Model 3 alignment probabilities , POS correspondence , and [[ bilingual dictionary coverage ]] as << features >> .
USED-FOR
[ 16, 18, 20, 20 ]
Our experiments show that [[ log-linear models ]] significantly outperform << IBM translation models >> .
COMPARE
[ 4, 5, 8, 10 ]
[[ Hough voting ]] in a geometric transformation space allows us to realize << spatial verification >> , but remains sensitive to feature detection errors because of the inflexible quan-tization of single feature correspondences .
USED-FOR
[ 0, 1, 11, 12 ]
<< Hough voting >> in a [[ geometric transformation space ]] allows us to realize spatial verification , but remains sensitive to feature detection errors because of the inflexible quan-tization of single feature correspondences .
FEATURE-OF
[ 4, 6, 0, 1 ]