Small Language Models can Outperform Humans in Short Creative Writing: A Study Comparing SLMs with Humans and LLMs

In this paper, we evaluate the creative fiction writing abilities of a fine-tuned small language model (SLM), BART Large, and compare its performance to humans and two large language models (LLMs): GPT-3.5 and GPT-4o. Our evaluation consists of two experiments: (i) a human evaluation where readers assess the stories generated by the SLM compared to human-written stories, and (ii) a qualitative linguistic analysis comparing the textual characteristics of the stories generated by the different models. In the first experiment, we asked 68 participants to rate short stories generated by the models and humans along dimensions such as grammaticality, relevance, creativity, and attractiveness. BART Large outperformed human writers in most aspects, except creativity, with an overall score of 2.11 compared to 1.85 for human-written texts -- a 14% improvement. In the second experiment, the qualitative analysis revealed that, while GPT-4o exhibited near-perfect internal and external coherence, it tended to produce more predictable narratives, with only 3% of its stories seen as novel. In contrast, 15% of BART's stories were considered novel, indicating a higher degree of creativity despite its smaller model size. This study provides both quantitative and qualitative insights into how model size and fine-tuning influence the balance between creativity, fluency, and coherence in creative writing tasks.

Pron vs Prompt: Can Large Language Models already Challenge a World-Class Fiction Author at Creative Text Writing?

It has become routine to report research results where Large Language Models (LLMs) outperform average humans in a wide range of language-related tasks, and creative text writing is no exception. It seems natural, then, to raise the bid: Are LLMs ready to compete in creative writing skills with a top (rather than average) novelist? To provide an initial answer for this question, we have carried out a contest between Patricio Pron (an awarded novelist, considered one of the best of his generation) and GPT-4 (one of the top performing LLMs), in the spirit of AI-human duels such as DeepBlue vs Kasparov and AlphaGo vs Lee Sidol. We asked Pron and GPT-4 to provide thirty titles each, and then to write short stories for both their titles and their opponent's. Then, we prepared an evaluation rubric inspired by Boden's definition of creativity, and we collected 5,400 manual assessments provided by literature critics and scholars. The results of our experimentation indicate that LLMs are still far from challenging a top human creative writer, and that reaching such level of autonomous creative writing skills probably cannot be reached simply with larger language models.

Automatic Scansion of Spanish Poetry without Syllabification

In recent years, several systems of automated metric analysis of Spanish poetry have emerged. These systems rely on complex methods of syllabification and stress assignment, which use PoS-tagging libraries, whose computational cost is high. This cost increases with the calculation of metric ambiguities. Furthermore, they do not consider determining issues in syllabic count such as the phenomena of compensation between hemistichs of verses of more than eleven syllables. However, it is possible to carry out an informative and accurate metric analysis without using these costly methods. We propose an algorithm that performs accurate scansion (number of syllables, stress pattern and type of verse) without syllabification. It addresses metric ambiguities and takes into account the hemistichs compensation. Our algorithm outperforms the current state of the art by 2% in fixed-metre poetry, and 25% in mixed-metre poetry. It also runs 21 and 25 times faster, respectively. Finally, a desktop application is offered as a tool for researchers of Spanish poetry.

An Effectiveness Metric for Ordinal Classification: Formal Properties and Experimental Results

In Ordinal Classification tasks, items have to be assigned to classes that have a relative ordering, such as positive, neutral, negative in sentiment analysis. Remarkably, the most popular evaluation metrics for ordinal classification tasks either ignore relevant information (for instance, precision/recall on each of the classes ignores their relative ordering) or assume additional information (for instance, Mean Average Error assumes absolute distances between classes). In this paper we propose a new metric for Ordinal Classification, Closeness Evaluation Measure, that is rooted on Measurement Theory and Information Theory. Our theoretical analysis and experimental results over both synthetic data and data from NLP shared tasks indicate that the proposed metric captures quality aspects from different traditional tasks simultaneously. In addition, it generalizes some popular classification (nominal scale) and error minimization (interval scale) metrics, depending on the measurement scale in which it is instantiated.

Combining Evaluation Metrics via the Unanimous Improvement Ratio and its Application to Clustering Tasks

Many Artificial Intelligence tasks cannot be evaluated with a single quality criterion and some sort of weighted combination is needed to provide system rankings. A problem of weighted combination measures is that slight changes in the relative weights may produce substantial changes in the system rankings. This paper introduces the Unanimous Improvement Ratio (UIR), a measure that complements standard metric combination criteria (such as van Rijsbergen's F-measure) and indicates how robust the measured differences are to changes in the relative weights of the individual metrics. UIR is meant to elucidate whether a perceived difference between two systems is an artifact of how individual metrics are weighted. Besides discussing the theoretical foundations of UIR, this paper presents empirical results that confirm the validity and usefulness of the metric for the Text Clustering problem, where there is a tradeoff between precision and recall based metrics and results are particularly sensitive to the weighting scheme used to combine them. Remarkably, our experiments show that UIR can be used as a predictor of how well differences between systems measured on a given test bed will also hold in a different test bed.

Towards Real-Time Summarization of Scheduled Events from Twitter Streams

This paper explores the real-time summarization of scheduled events such as soccer games from torrential flows of Twitter streams. We propose and evaluate an approach that substantially shrinks the stream of tweets in real-time, and consists of two steps: (i) sub-event detection, which determines if something new has occurred, and (ii) tweet selection, which picks a representative tweet to describe each sub-event. We compare the summaries generated in three languages for all the soccer games in "Copa America 2011" to reference live reports offered by Yahoo! Sports journalists. We show that simple text analysis methods which do not involve external knowledge lead to summaries that cover 84% of the sub-events on average, and 100% of key types of sub-events (such as goals in soccer). Our approach should be straightforwardly applicable to other kinds of scheduled events such as other sports, award ceremonies, keynote talks, TV shows, etc.

The Uned systems at Senseval-2

We have participated in the SENSEVAL-2 English tasks (all words and lexical sample) with an unsupervised system based on mutual information measured over a large corpus (277 million words) and some additional heuristics. A supervised extension of the system was also presented to the lexical sample task. Our system scored first among unsupervised systems in both tasks: 56.9% recall in all words, 40.2% in lexical sample. This is slightly worse than the first sense heuristic for all words and 3.6% better for the lexical sample, a strong indication that unsupervised Word Sense Disambiguation remains being a strong challenge.

The Role of Conceptual Relations in Word Sense Disambiguation

We explore many ways of using conceptual distance measures in Word Sense Disambiguation, starting with the Agirre-Rigau conceptual density measure. We use a generalized form of this measure, introducing many (parameterized) refinements and performing an exhaustive evaluation of all meaningful combinations. We finally obtain a 42% improvement over the original algorithm, and show that measures of conceptual distance are not worse indicators for sense disambiguation than measures based on word-coocurrence (exemplified by the Lesk algorithm). Our results, however, reinforce the idea that only a combination of different sources of knowledge might eventually lead to accurate word sense disambiguation.

Indexing with WordNet synsets can improve Text Retrieval

The classical, vector space model for text retrieval is shown to give better results (up to 29% better in our experiments) if WordNet synsets are chosen as the indexing space, instead of word forms. This result is obtained for a manually disambiguated test collection (of queries and documents) derived from the Semcor semantic concordance. The sensitivity of retrieval performance to (automatic) disambiguation errors when indexing documents is also measured. Finally, it is observed that if queries are not disambiguated, indexing by synsets performs (at best) only as good as standard word indexing.

Generic rules and non-constituent coordination

We present a metagrammatical formalism, {\em generic rules}, to give a default interpretation to grammar rules. Our formalism introduces a process of {\em dynamic binding} interfacing the level of pure grammatical knowledge representation and the parsing level. We present an approach to non-constituent coordination within categorial grammars, and reformulate it as a generic rule. This reformulation is context-free parsable and reduces drastically the search space associated to the parsing task for such phenomena.