1,720,996 research outputs found
Zero-Shot Metrical Poetry Generation with Open Language Models: a Quantitative Analysis
No full textPoetry generation, a long endeavor in Computational Creativity, can nowadays be tackled by prompting an LLM. From a set of natural language instructions, which may cover semantic and formal constraints, the result is generally a piece of text that one would recognize as poetry, but where metrical constraints are not always met. To confirm this empirically and take conclusions on its extent, we measure how well generated poems match simple constraints such as the number of lines, stanzas, or syllables per line. Specifically, we prompt the open LLM Llama2 with instructions of variable complexity; we compute the previous constraints in generated poems using an automatic tool; and we analyse how much they deviate from the given instructions. We conclude that, in a simple prompting scenario, the general purpose Llama2 model is unlikely to produce well-metered text
Toward Multimodal Sentiment Analysis of Historic Plays: A Case Study with Text and Audio for Lessing’s Emilia Galotti
Full text linkWe present a case study as part of a work-in-progress project about multimodal sentiment analysis on historic German plays, taking Emilia Galotti by G. E. Lessing as our initial use case. We analyze the textual version and an audio version (audiobook). We focus on ready-to-use sentiment analysis methods: For the textual component, we implement a naive lexicon-based approach and another approach that enhances the lexicon by means of several NLP methods. For the audio analysis, we use the free version of the Vokaturi tool. We compare the results of all approaches and evaluate them against the annotations of a human expert, which serves as a gold standard. For our use case, we can show that audio and text sentiment analysis behave very differently: textual sentiment analysis tends to predict sentiment as rather negative and audio sentiment as rather positive. Compared to the gold standard, the textual sentiment analysis achieves accuracies of 56% while the accuracy for audio sentiment analysis is only 32%. We discuss possible reasons for these mediocre results and give an outlook on further steps we want to pursue in the context of multimodal sentiment analysis on historic plays
Exploring the Distinction: Investigating the Recognition of Automatic Text Generation Systems and Differentiating Human Text from Language Models.
No full textKUCST at CheckThat 2023:How good can we be with a generic model?
No full textIn this paper we present our method for tasks 2 and 3A at the CheckThat2023 shared task. We make use of a generic approach that has been used to tackle a diverse set of tasks, inspired by authorship attribution and profiling. We train a number of Machine Learning models and our results show that Gradient Boosting performs the best for both tasks. Based on the official ranking provided by the shared task organizers, our model shows an average performance compared to other teams
Creating an Aligned Corpus of Sound and Text: The Multimodal Corpus of Shakespeare and Milton
Full text linkIn this work we present a corpus of poems by William Shakespeare and John Milton that have been enriched with readings from the public domain. We have aligned all the lines with their respective audio segments, at the line, word, syllable and phone level, and we have included their scansion. We make a basic visualization platform for these poems and we conclude by conjecturing possible future directions
Recommended from our members
A Humorous View into the Past: The Old Jokes Archive
Full text linkJokes represent one of the most understudied sources about nineteenth century society. Due to their ephemeral nature they slipped from attention as soon as they were no longer funny or topical. Digitisation of newspapers and books has made them available again, but due to their short nature they are not easily accessible through current generic keyword-based newspaper search systems. In this paper we present the Old Jokes Archive, which aims to provide a digital archive focused solely on jokes. The archive will support the full process from initial text acquisition to search and finally re-use by both academic and general public users
Conservatism in an Innovative Field : Childrens Digital Books in Sweden
No full textThe study intention is to understand the nature of digital publishing for childrenin Sweden and showed, among other things, that digital publishing to a large extentis reproductions of print books. The paper argues that results of the case studycan be explained by long-time structures in publishing of print books for childrenbut also that there are barrier breakers.In order to map digital book publishing for children data was gathered from anumber of sources and covered digital children’s books published in Swedish2015–2018. The preliminary results of the study points towards severe gaps inthe statistics of children’s digital books and digital publishing in general. Despitethe lack of exact statistics observations can be made on the nature of the digitalchildren’s book market in terms of publishing formats, development, publishersand distribution.The results of this study showed a print-based publishing structure of digitalbooks. I argue that this contradiction can be explained by five factors: publishingas a business, copyright, production, authors and audience. These five are linkedbut provide different perspectives and explanations
Diane Simmons at SemEval-2023 Task 5:Is it possible to make good clickbait spoilers using a Zero-Shot approach? Check it out!
No full text- …
