arXiv subjects
Publications and source records attributed to Taylor Arnold.
This study investigates silence gaps in two kinds of audiovisual material. We analysed thirty US situational comedies and fifty-one synthetic podcasts generated with Google NotebookLM. Gaps were compared across speaker gender, assigned from a fundamental-frequency threshold estimated in Praat, and across production settings.
Modern vision-language models (VLMs) often fail at cultural competency evaluations and benchmarks. Given the diversity of applications built upon VLMs, there is renewed interest in understanding how they encode cultural nuances. While individual aspects of this problem have been studied, we still lack a comprehensive framework for systematically identifying and annotating the nuanced cultural dimensions present in images for VLMs. This position paper argues that foundational methodologies from visual culture studies (cultural studies, semiotics, and visual studies) are necessary for cultural analysis of images. Building upon this review, we propose a set of five frameworks, corresponding to cultural dimensions, that must be considered for a more complete analysis of the cultural competencies of VLMs.
In the 1970s, the United States Environmental Protection Agency sponsored Documerica, a large-scale photography initiative to document environmental subjects nation-wide. While over 15,000 digitized public-domain photographs from the collection are available online, most of the images were scanned from damaged copies of the original prints. We present and evaluate a modified histogram matching technique based on the underlying chemistry of the prints for correcting the damaged images by using training data collected from a small set of undamaged prints. The entire set of color-adjusted Documerica images is made available in an open repository.
Many cultural institutions have made large digitized visual collections available online, often under permissible re-use licences. Creating interfaces for exploring and searching these collections is difficult, particularly in the absence of granular metadata. In this paper, we introduce a method for using state-of-the-art multimodal large language models (LLMs) to enable an open-ended, explainable search and discovery interface for visual collections. We show how our approach can create novel clustering and recommendation systems that avoid common pitfalls of methods based directly on visual embeddings. Of particular interest is the ability to offer concrete textual explanations of each recommendation without the need to preselect the features of interest. Together, these features can create a digital interface that is more open-ended and flexible while also being better suited to addressing privacy and ethical concerns. Through a case study using a collection of documentary photographs, we provide several metrics showing the efficacy and possibilities of our approach.
Museums, libraries, and other cultural institutions continue to prioritize and build web-based visualization systems that increase access and discovery to digitized archives. Prominent examples exist that illustrate impressive visualizations of a particular feature of a collection. For example, interactive maps showing geographic spread or timelines capturing the temporal aspects of collections. By way of a case study, this paper presents a new web-based visualization system that allows users to simultaneously explore a large collection of images along several different dimensions---spatial, temporal, visual, textual, and through additional metadata fields including the photographer name---guided by the concept of generous interfaces. The case study is a complete redesign of a previously released digital, public humanities project called Photogrammar (2014). The paper highlights the redesign's interactive visualizations that are now possible by the affordances of newly available software. All of the code is open-source in order to allow for re-use of the codebase to other collections with a similar structure.
This paper analyses the contribution of language metrics and, potentially, of linguistic structures, to classify French learners of English according to levels of the Common European Framework of Reference for Languages (CEFRL). The purpose is to build a model for the prediction of learner levels as a function of language complexity features. We used the EFCAMDAT corpus, a database of one million written assignments by learners. After applying language complexity metrics on the texts, we built a representation matching the language metrics of the texts to their assigned CEFRL levels. Lexical and syntactic metrics were computed with LCA, LSA, and koRpus. Several supervised learning models were built by using Gradient Boosted Trees and Keras Neural Network methods and by contrasting pairs of CEFRL levels. Results show that it is possible to implement pairwise distinctions, especially for levels ranging from A1 to B1 (A1=>A2: 0.916 AUC and A2=>B1: 0.904 AUC). Model explanation reveals significant linguistic features for the predictiveness in the corpus. Word tokens and word types appear to play a significant role in determining levels. This shows that levels are highly dependent on specific semantic profiles.
Word choice is dependent on the cultural context of writers and their subjects. Different words are used to describe similar actions, objects, and features based on factors such as class, race, gender, geography and political affinity. Exploratory techniques based on locating and counting words may, therefore, lead to conclusions that reinforce culturally inflected boundaries. We offer a new method, the DualNeighbors algorithm, for linking thematically similar documents both within and across discursive and linguistic barriers to reveal cross-cultural connections. Qualitative and quantitative evaluations of this technique are shown as applied to two cultural datasets of interest to researchers across the humanities and social sciences. An open-source implementation of the DualNeighbors algorithm is provided to assist in its application.
The package cleanNLP provides a set of fast tools for converting a textual corpus into a set of normalized tables. The underlying natural language processing pipeline utilizes Stanford's CoreNLP library, exposing a number of annotation tasks for text written in English, French, German, and Spanish. Annotators include tokenization, part of speech tagging, named entity recognition, entity linking, sentiment analysis, dependency parsing, coreference resolution, and information extraction.
The iotools package provides a set of tools for Input/Output (I/O) intensive datasets processing in R (R Core Team, 2014). Efficent parsing methods are included which minimize copying and avoid the use of intermediate string representations whenever possible. Functions for applying chunk-wise operations allow for computing on streaming input as well as arbitrarily large files. We present a set of example use cases for iotools, as well as extensive benchmarks comparing comparable functions provided in both core-R as well as other contributed packages.
Large spatial datasets often represent a number of spatial point processes generated by distinct entities or classes of events. When crossed with covariates, such as discrete time buckets, this can quickly result in a data set with millions of individual density estimates. Applications that require simultaneous access to a substantial subset of these estimates become resource constrained when densities are stored in complex and incompatible formats. We present a method for representing spatial densities along the nodes of sparsely populated trees. Fast algorithms are provided for performing set operations and queries on the resulting compact tree structures. The speed and simplicity of the approach is demonstrated on both real and simulated spatial data.
We present a modification of the standard geohash algorithm based on maximum entropy encoding in which the data volume is approximately constant for a given hash prefix length. Distributed spatiotemporal databases, which typically require interleaving spatial and temporal elements into a single key, reap large benefits from a balanced geohash by creating a consistent ratio between spatial and temporal precision even across areas of varying data density. This property is also useful for indexing purely spatial datasets, where the load distribution of large range scans is an important aspect of query performance. We apply our algorithm to data generated proportional to population as given by census block population counts provided from the US Census Bureau.
We consider efficient implementations of the generalized lasso dual path algorithm of Tibshirani and Taylor (2011). We first describe a generic approach that covers any penalty matrix D and any (full column rank) matrix X of predictor variables. We then describe fast implementations for the special cases of trend filtering problems, fused lasso problems, and sparse fused lasso problems, both with X=I and a general matrix X. These specialized implementations offer a considerable improvement over the generic implementation, both in terms of numerical stability and efficiency of the solution path computation. These algorithms are all available for use in the genlasso R package, which can be found in the CRAN repository.