Jenny Rose Finkel's research while affiliated with Stanford University and other places

Publications (23)

Conference Paper
Full-text available
We describe the design and use of the Stanford CoreNLP toolkit, an extensible pipeline that provides core natural language analysis. This toolkit is quite widely used, both in the research NLP community and also among commercial and government users of open source NLP technology. We suggest that this follows from a simple, approachable design, stra...
Article
Full-text available
This document describes Stanford University’s first entry into a NIST MT evaluation. Our entry to the 2008 evaluation mainly focused on establishing a competent baseline with a phrase-based system similar
Conference Paper
Full-text available
One of the main obstacles to produc- ing high quality joint models is the lack of jointly annotated data. Joint model- ing of multiple natural language process- ing tasks outperforms single-task models learned from the same data, but still under- performs compared to single-task models learned on the more abundant quantities of available single-tas...
Conference Paper
Full-text available
For many language technology applications, such as question answering, the overall sys- tem runs several independent processors over the data (such as a named entity recognizer, a coreference system, and a parser). This eas- ily results in inconsistent annotations, which are harmful to the performance of the aggre- gate system. We begin to address...
Conference Paper
Full-text available
Many named entities contain other named entities inside them. Despite this fact, the field of named entity recognition has al- most entirely ignored nested named en- tity recognition, but due to technological, rather than ideological reasons. In this pa- per, we present a new technique for rec- ognizing nested named entities, by using a discriminat...
Conference Paper
Full-text available
Multi-task learning is the problem of maxi- mizing the performance of a system across a number of related tasks. When applied to mul- tiple domains for the same task, it is similar to domain adaptation, but symmetric, rather than limited to improving performance on a target domain. We present a more principled, better performing model for this prob...
Conference Paper
Full-text available
Discriminative feature-based methods are widely used in natural language processing, but sentence parsing is still dominated by generative methods. While prior feature-based dynamic programming parsers have restricted training and evaluation to artificially short sentences, we present the first general, featurerich discriminative parser, based on a...
Conference Paper
Full-text available
A desirable quality of a coreference resolution system is the ability to handle transitivity con- straints, such that even if it places high like- lihood on a particular mention being corefer- ent with each of two other mentions, it will also consider the likelihood of those two men- tions being coreferent when making a final as- signment. This is...
Conference Paper
Full-text available
Historically, unsupervised learning tech- niques have lacked a principled technique for selecting the number of unseen compo- nents. Research into non-parametric priors, such as the Dirichlet process, has enabled in- stead the use of infinite models, in which the number of hidden categories is not fixed, but can grow with the amount of training dat...
Conference Paper
The end-to-end performance of natural language processing systems for compound tasks, such as question answering and textual entailment, is often hampered by use of a greedy 1-best pipeline architecture, which causes errors to propagate and compound at each stage. We present a novel architecture, which models these pipelines as Bayesian networks, w...
Conference Paper
Full-text available
The end-to-end performance of natural language processing systems for com- pound tasks, such as question answering and textual entailment, is often hampered by use of a greedy 1-best pipeline archi- tecture, which causes errors to propagate and compound at each stage. We present a novel architecture, which models these pipelines as Bayesian network...
Article
Full-text available
We present a machine learning approach to ro- bust textual inference, in which parses of the text and the hypothesis sentences are used to mea- sure their asymmetric "similarity", and thereby to decide if the hypothesis can be inferred. This idea is realized in two different ways. In the first, each sentence is represented as a graph (extracted fro...
Article
Full-text available
Good automatic information extraction tools offer hope for automatic processing of the exploding biomedical literature, and successful named entity recognition is a key component for such tools. We present a maximum-entropy based system incorporating a diverse set of features for identifying gene and protein names in biomedical abstracts. This syst...
Article
Full-text available
We present a maximum entropy-based system for identifying named entities (NEs) in biomedical abstracts and present its performance in the only two biomedical named entity recognition (NER) comparative evaluations that have been held to date, namely BioCreative and Coling BioNLP. Our system obtained an exact match F-score of 83.2% in the BioCreative...
Conference Paper
Full-text available
Most current statistical natural language process- ing models use only local features so as to permit dynamic programming in inference, but this makes them unable to fully account for the long distance structure that is prevalent in language use. We show how to solve this dilemma with Gibbs sam- pling, a simple Monte Carlo method used to per- form...
Article
Full-text available
We describe a machine learning system for the recognition of names in biomedical texts. The system makes extensive use of local and syntactic features within the text, as well as external resources including the web and gazetteers. It achieves an F-score of 70% on the Coling 2004 NLPBA/BioNLP shared task of identifying five biomedical named entitie...
Conference Paper
Full-text available
We describe a machine learning system for the recognition of names in biomedical texts. The system makes extensive use of local and syntactic features within the text, as well as external resources including the web and gazetteers. It achieves an F-score of 70% on the Coling 2004 NLPBA/BioNLP shared task of identifying five biomedical named entitie...
Article
This paper demonstrates how genetic programming can be used to evolve an algorithm to correctly factor positive integers.
Article
Full-text available
We initially describe a feature-rich discriminative Conditional Random Field (CRF) model for Infor-mation Extraction in the workshop announcements domain, which offers good baseline performance in the PASCAL shared task. We then propose a method for leveraging domain knowledge in Infor-mation Extraction tasks, scoring candidate docu-ment labellings...
Article
Full-text available
This document describes Stanford University’s first entry into a NIST MT evaluation. Our entry to the 2008 evaluation mainly focused on establishing a competent baseline with a phrase-based system similar to (Och and Ney, 2004; Koehn et al., 2007). In a three-week effort prior to the evaluation, our attention focused on scaling up our system to exp...

Citations

... In line with the findings of Son et al. (2020), which state that neutral sentiments represent consumers' expectations, we conducted sentiment analysis to validate H1-1 and H1-2 by utilizing the (Manning et al., 2014), a natural language processing toolkit. The sentiment analyzer produces positive, neutral, and negative sentiment proportions per document, which are summed to 1.0 (Socher et al., 2013). ...
... Traditional event tasks mostly contained classifiers based on pattern recognition or machine learning methods, such as Monte Carlo Gibbs sampling [9], conditional random fields [10], support vector machines [11], and so on. With the extensive application of deep learning, deep neural network model is also more and more applied to the task of event extraction, such as convolutional neural network [12] and graph neural network [13]. ...
... (This is not relevant as gene is the only entity type we are concerned with.) The first and second steps are addressed in task 1A (Finkel et al., 2004). Third, the entities are grounded with respect to its denotation in the world (or model of the world). ...
... Collier et al. (2000) and Koichi and Collier (2003) attempt a 10-NE task, using a private corpus to evaluate, and report Fscores of 74 and 73. We have analysed our sources of error for both BioCreative and BioNLP in depth in Dingare et al. (2004) and Finkel et al. (2004); these include a large percentage of boundary errors (over 30% for both tasks), a smaller number of errors due to coordination, and some errors due to acronyms and tokens, whose orthographic form might suggest they were entities but were in fact measures or belonged to other entity categories; also a number of errors due to low-frequency words or words not encountered in the training data. However, we would like to focus here on the quality of training and evaluation data as a key factor leading to low performance. ...
... However, these models cannot learn rewrite rules such as 'the X VERB Y X does not VERB Y ', which are instead learned by our model. Although, several systems tried to leverage large repositories such as DIRT (with limited success; de Salvo Braz et al. 2005b;Raina et al. 2005), the combined use of the two forms of extracting first-order rewrite rules is a very interesting research line. Pilot experiments using verbs in entailment extracted with the method presented in Zanzotto, Pennacchiotti and Pazienza (2006) and our model have shown promising results. ...
... Pioneering work in terms of loops in GP has been done by Koza [4]. In this paragraph, we outline the contributions of two of his students: A new syntax in which conditional loops and alternatives were used was established by Finkel [3]. She solved the factoring problem with this approach and applied a penalty in the fitness to keep the programs small. ...
... Luoma and Pyysalo (2020) showed that context information from neighboring sentences has positive effects for named entity recognition on the general domain. (Finkel et al., 2004) also showed the positive impact of context for clinical concept extraction. We follow these approaches and add context information to the input similar to (Schweter and Akbik, 2020). ...
... Information extraction Machine learning Date, CFP, Topic, Sponsor and Program committee Xin et al. [24] Title, Date, Deadline, Location, Name, Title and URL Schneider [19] Title, Date, URL and Acronym Cox et al. [5] Title, Date, Deadline, Location, Homepage and Acronym Kim et al. [8] Title, Date, Deadline, Location, URL and Contact number Eom [6] Title, Date, Deadline, Location, Homepage and Acronym Li et al. [14] Speaker, Location and Time Ciravegna [2] Information retrieval Similarity between document and query Date and Country Lazarinis [10] <Table div, p, ul, li, h1, h2, h3, h4, h5, h6, table Non-informative attribute list combx, comment, com-, contact, foot, footer, footnote, masthead, media, meta, outbrain, promo, related, scroll, shoutbox, sidebar, sponsor, shopping, tags, tool, widget, menu, fax, download, register, admin, copy. ...
... NLP is a pipeline processing because it consists of many processing stages in series and the subsequent stages are dependent on the output from previous stage. A wellknown drawback of this pipeline processing is cascading error propagation caused by the residual error that is generated by each stage and affects overall performance of NLP [5]. Hence, each stage in the pipeline process must achieve high accuracy and less errors to get good overall performance. ...
... A few previous works have designed proprietary structures to deal with the nested entities, such as the constituency graph [9] and hypergraph [10]. Other works [11,12] capture entities through the layered model containing multiple recognition layers. ...