Dominican Republic

A common practice in building NLP datasets, especially using crowd-sourced annotations, involves obtaining multiple annotator judgements on the same data instances, which are then flattened to produce a single ``ground truth&#39;&#39; label or score, through majority voting, averaging, or adjudication. While these approaches may be appropriate in certain annotation tasks, such aggregations overlook the socially constructed nature of human perceptions that annotations for relatively more subjective tasks are meant to capture. In particular, systematic disagreements between annotators owing to their socio-cultural backgrounds and/or lived experiences are often obfuscated through such aggregations. In this work, we empirically demonstrate that label aggregation may introduce representational biases of individual and group perspectives. Based on this finding, we propose a set of recommendations for increased utility and transparency of datasets for downstream use cases.

EMNLP 2021

On Releasing Annotator-Level Labels and Information in Datasets

EMNLP 2021 is planned to be a hybrid event in Punta Cana, Dominican Republic, with both on-site and fully virtual participation possible. The experience for on-site participants would closely approximate a normal pre-COVID *ACL conference, with 5-6 thematically organized parallel sessions and live Q/A and interactive discussion immediately after the talks. Presentations by virtual participants will be equitably interleaved with those of on-site participants, projected on the auditorium screens as if on-site, and also followed immediately by live Q/A and interactive discussion at a time during reasonable waking hours for the virtual presenter. For all participants, on-site and virtual, who are unable to attend a session due to either time-zone issues or because they are participating in another session live, talk recordings and slides will be available online at a minimum after the live presentation (and in many cases before as well), and questions may be submitted in advance on session-specific discussion boards and answered live in session with the usual visual aids if desired.

<iframe style="width:700px;height:400px" src="https://online.fliphtml5.com/ebtyf/ceby/"  seamless="seamless" scrolling="no" frameborder="0" allowtransparency="true" allowfullscreen="true" ></iframe>

Please Note: The EMNLP registration system is not currently connected to the underline site as we are still in the process of building out EMNLP 2021. You will receive access instructions from underline the week of November 1st. 

Access is given only to EMNLP upon registration, if you have not registered please do so [here](https://2021.emnlp.org/registration).

Registered attendees will receive access the week of November 1st.

EMNLP 2021 is planned to be a hybrid event in Punta Cana, Dominican Republic, with both on-site and fully virtual participation possible.

A common practice in building NLP datasets, especially using crowd-sourced annotations, involves obtaining multiple annotator judgements on the same data instances, which are then flattened to produce a single ``ground truth'' label or score, through majority voting, averaging, or adjudication. While these approaches may be appropriate in certain annotation tasks, such aggregations overlook the socially constructed nature of human perceptions that annotations for relatively more subjective tasks are meant to capture. In particular, systematic disagreements between annotators owing to their socio-cultural backgrounds and/or lived experiences are often obfuscated through such aggregations. In this work, we empirically demonstrate that label aggregation may introduce representational biases of individual and group perspectives. Based on this finding, we propose a set of recommendations for increased utility and transparency of datasets for downstream use cases.

workshop paper

With the recent rise in popularity of Transformer models in Natural Language Processing, research efforts have been dedicated to the development of domain-adapted versions of BERT-like architectures. In this study, we focus on FinBERT, a Transformer model trained on text from the financial domain. By comparing its performances with the original BERT on a wide variety of financial text processing tasks, we found continual pretraining from the original model to be the more beneficial option. Domain-specific pretraining from scratch, conversely, seems to be less effective

Is Domain Adaptation Worth Your Investment? Comparing BERT and FinBERT on Financial Tasks

In decision making in the economic field, an especially important requirement is to rapidly understand news to absorb ever-changing economic situations. Given that most economic news is written in English, the ability to read such information without waiting for a translation is particularly valuable in economics in contrast to other fields. In consideration of this issue, this research investigated the extent to which non-native English speakers are able to read economic news to make decisions accordingly -- an issue that has been rarely addressed in previous studies. Using an existing standard dataset as training data, we created a classifier that automatically evaluates the readability of text with high accuracy for English learners. Our assessment of the readability of an economic news corpus revealed that most news texts can be read by intermediate English learners. We also found that in some cases, readability varies considerably depending on the knowledge of certain words specific to the economic field.

To What Extent Can English-as-a-Second Language Learners Read Economic News Texts?

We present a method to automatically identify financially relevant news using stock price movements and news headlines as input. The method repurposes the attention weights of a neural network initially trained to predict stock prices to assign a relevance score to each headline, eliminating the need for manually labeled training data. Our experiments on the four most relevant US stock indices and 1.5M news headlines show that the method ranks relevant news highly, positively correlated with the accuracy of the initial stock price prediction task.

From Stock Prediction to Financial Relevance: Repurposing Attention Weights to Assess News Relevance Without Manual Annotations

In this paper we investigate the possibility of extracting predicate-argument relations from UD trees (and enhanced UD graphs). Concretely, we apply UD parsers on an English question answering/semantic-role labeling data set (FitzGerald, 2018) and check if the annotations reflect the relations in the resulting parse trees, using a small number of rules to extract this information. We find that 79.1\% of the argument-predicate pairs can be found in this way, on the basis of Udify (Kondratyuk and Straka, 2019). Error analysis reveals that half of the error cases are attributable to shortcomings in the dataset. The remaining errors are mostly due to predicate-argument relations not being extractible algorithmically from the UD trees (requiring semantic reasoning to be resolved). The parser itself is only responsible for a small portion of errors. Our analysis suggests a number of improvements to the UD annotation schema: we propose to enhance the schema in four ways, in order to capture argument-predicate relations. Additionally, we propose improvements regarding data collection for question answering/semantic-role labeling data.

Can predicate-argument relationships be extracted from UD trees?

While FrameNet is widely regarded as a rich resource of semantics in natural language processing, a major criticism concerns its lack of coverage and the relative paucity of its labeled data compared to other commonly used lexical resources such as PropBank and VerbNet. This paper reports on a pilot study to address these gaps. We propose a data augmentation approach, which uses existing frame-specific annotation to automatically annotate other lexical units of the same frame which are unannotated. Our rule-based approach defines the notion of a sister lexical unit and generates frame-specific augmented data for training. We present experiments on frame-semantic role labeling which demonstrate the importance of this data augmentation: we obtain a large improvement to prior results on frame identification and argument identification for FrameNet, utilizing both full-text and lexicographic annotations under FrameNet. Our findings on data augmentation highlight the value of automatic resource creation for improved models in frame-semantic parsing.

Sister Help: Data Augmentation for Frame-Semantic Role Labeling

In this paper, we present a first attempt at enriching German Universal Dependencies (UD) treebanks with enhanced dependencies. Similarly to the converter for English (Schuster and Manning, 2016), we develop a rule-based system for deriving enhanced dependencies from the basic layer, covering three linguistic phenomena: relative clauses, coordination, and raising/control. For quality control, we manually correct or validate a set of 196 sentences, finding that around 90% of added relations are correct. Our data analysis reveals that difficulties arise mainly due to inconsistencies in the basic layer annotations. We show that the English system is in general applicable to German data, but that adapting to the particularities of the German treebanks and language increases precision and recall by up to 10%. Comparing the application of our converter on gold standard dependencies vs. automatic parses, we find that F1 drops by around 10% in the latter setting due to error propagation. Finally, an enhanced UD parser trained on a converted treebank performs poorly when evaluated against our annotations, indicating that more work remains to be done to create gold standard enhanced German treebanks.

A Corpus Study of Creating Rule-Based Enhanced Universal Dependencies for German

Universal Conceptual Cognitive Annotation (UCCA) is a semantic annotation scheme that organizes texts into coarse predicate-argument structure, offering broad coverage of semantic phenomena. At the same time, there is still need for a finer-grained treatment of many of the categories. The Adverbial category is of special interest, as it covers a wide range of fundamentally different meanings such as negation, causation, aspect, and event quantification. In this paper we introduce a refinement annotation scheme for UCCA's Adverbial category, showing that UCCA Adverbials can indeed be subcategorized into at least 7 semantic types, and doing so can help clarify and disambiguate the otherwise coarse-grained labels. We provide a preliminary set of annotation guidelines, as well as pilot annotation experiments with high inter-annotator agreement, confirming the validity of the scheme.

Subcategorizing Adverbials in Universal Conceptual Cognitive Annotation

While annotating normalized times in food security documents, we found that the semantically compositional annotation for time normalization (SCATE) scheme required several near-duplicate annotations to get the correct semantics for expressions like Nov. 7th to 11th 2021.
To reduce this problem, we explored replacing SCATE's Sub-Interval property with a Super-Interval property, that is, making the smallest units (e.g., 7th and 11th) rather than the largest units (e.g., 2021) the heads of the intersection chains.
To ensure that the semantics of annotated time intervals remained unaltered despite our changes to the syntax of the annotation scheme, we applied several different techniques to validate our changes.
These validation techniques detected and allowed us to resolve several important bugs in our automated translation from Sub-Interval to Super-Interval syntax.

Simplifying annotation of intersections in time normalization annotation: exploring syntactic and semantic validation

This paper presents several challenges faced when annotating Turkish treebanks in accordance with the Universal Dependencies (UD) guidelines and proposes solutions to address them. Most of these challenges stem from the lack of adequate support in the UD framework to accurately represent null morphemes and complex derivations, which results in a significant loss of information for Turkish. This loss negatively impacts the tools that are developed based on these treebanks. 
We raised and discussed these issues within the community on the official UD portal. This paper presents these issues and our proposals to more accurately represent morphosyntactic information for Turkish while adhering to guidelines of UD. This work aims to contribute to the representation of Turkish and other agglutinative languages in UD-based treebanks, which in turn aids to develop more accurately annotated datasets for such languages.

Overcoming the challenges in morphological annotation of Turkish in universal dependencies framework

Using automatically generated VerbNet semantic representations, we extract events and their participants and predict the changes in existence and location states of entities. The results are evaluated on the ProPara dataset.

On Releasing Annotator-Level Labels and Information in Datasets

Next from EMNLP 2021

Is Domain Adaptation Worth Your Investment? Comparing BERT and FinBERT on Financial Tasks

Similar lecture

Documenting Large Webtext Corpora: A Case Study on the Colossal Clean Crawled Corpus