pith. sign in

arxiv: 2604.19111 · v1 · submitted 2026-04-21 · 💻 cs.HC

Revisiting Framing Codebooks with AI: Employing Large Language Models as Analytical Collaborators in Deductive Content Analysis

Diego Gomez-Zara , Hern\'an Valdivieso , Jorge P\'erez , Denis Parra , Sebasti\'an Valenzuela This is my paper

Pith reviewed 2026-05-10 02:30 UTC · model grok-4.3

classification 💻 cs.HC
keywords framing analysiscodebookslarge language modelsdeductive content analysisqualitative methodsnews mediaAI collaborationiterative refinement
0
0 comments X

The pith

Large language models can act as analytic collaborators to externalize decision rules and refine framing codebooks through iterative researcher dialogues.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proposes a workflow that treats large language models as partners rather than automated classifiers when building and updating codebooks for deductive framing analysis. Codebooks start from theoretical frameworks but often reveal ambiguities and gaps when applied to actual news data, especially as coverage shifts across time and cultures. By engaging LLMs in dialogues, researchers can make their coding rules explicit, uncover latent dimensions in the data, and revise the codebooks iteratively while retaining final interpretive control. The approach is illustrated with Latin American news coverage, where it helped surface new frame distinctions and adapt existing frameworks to local contexts. The result is a method that supports methodological creativity without replacing human judgment.

Core claim

LLMs can augment the creation and refinement of framing codebooks by combining theoretical frameworks with data-driven exploration, where the models serve as analytic collaborators that externalize decision rules, surface latent dimensions, and support iterative revisions through dialogues between researchers and their data, as shown in an application to Latin American news coverage that generated frame distinctions and adapted frameworks to new contexts.

What carries the argument

The LLM-assisted iterative workflow for codebook refinement, in which models participate in researcher dialogues to externalize rules and identify latent patterns while researchers retain interpretive authority.

If this is right

  • Codebooks can adapt more readily to evolving news corpora and cross-cultural differences without requiring complete redevelopment from theory.
  • Latent framing dimensions that theory alone overlooks become visible through data-driven dialogue with the models.
  • Researchers maintain full control over final codebook decisions while gaining structured support for exploring ambiguities.
  • The workflow extends to other deductive content analysis tasks where initial theoretical criteria require data-grounded sharpening.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The method could enable more responsive codebooks that update continuously as new news data arrives.
  • It suggests a broader pattern for integrating AI tools into qualitative research while keeping human oversight central.
  • Validation experiments comparing LLM-assisted codebooks against purely manual ones on held-out datasets would test the approach's reliability across domains.
  • The technique may reduce initial development time for codebooks but shifts effort toward ongoing dialogue and bias-checking steps.

Load-bearing premise

Large language models can reliably externalize decision rules, surface latent patterns, and support valid iterative revisions without introducing systematic biases, hallucinations, or interpretive distortions that undermine the codebooks' theoretical integrity.

What would settle it

A direct comparison on the same news corpus where frames identified by the LLM-refined codebook systematically differ from those identified by an independently developed traditional codebook in ways that match documented model biases or omissions.

Figures

Figures reproduced from arXiv: 2604.19111 by Denis Parra, Diego Gomez-Zara, Hern\'an Valdivieso, Jorge P\'erez, Sebasti\'an Valenzuela.

Figure 1
Figure 1. Figure 1: Overview of the LLM Codebook Prompt frame definitions, researchers prompt the LLM to tentatively identify the prevalence of these frames in the corpus, uncover potential frames, and flag uncertain cases. By interacting with the LLM, researchers can request summaries of recurring rationales and identify clusters that merit closer examination. For example, using Iyengar’s thematic framing framework for polit… view at source ↗
Figure 2
Figure 2. Figure 2: Prompt used for frame classification. 3.5 Phase 5. Analytic Interrogation and Criteria Elicitation Using the initial codebook, researchers systematically interrogate the LLM’s explanations to surface implicit decision rules, latent criteria, and classification inconsistencies. Rather than evaluating outputs solely for accuracy, the goal is to examine how the LLM interprets the codebook and where its reason… view at source ↗
Figure 3
Figure 3. Figure 3: LLM’s response to the Initial Prompt. (e.g., “The government is responsible...,” “... authorities failed...”). The least frequent was ‘morality,’ which the LLM noted “is rarely explicit; it usually appears diluted within the conflict or responsibility.” According to the LLM, this frame was related to normative language (e.g., “outrageous,” “unacceptable”), ethical judgments, and discussions on religion and… view at source ↗
Figure 4
Figure 4. Figure 4: Initial LLM Codebook Prompt employed in Valenzuela et al.’s study (2017) [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Ambiguous case in the Morality Frame. Source: Valenzuela et al. (2023), ID: 25801. [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Final LLM Codebook Prompt employed in Valenzuela et al.’s study (2017) [PITH_FULL_IMAGE:figures/full_fig_p015_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: (Continued) Final LLM Codebook Prompt employed in Valenzuela et al.’s study [PITH_FULL_IMAGE:figures/full_fig_p016_7.png] view at source ↗
read the original abstract

Codebooks are central to framing research, providing theoretically grounded criteria for analyzing news content. While traditionally codebooks are built from theoretical frameworks and researchers' knowledge, applying these codebooks to large news corpora often exposes ambiguities, borderline cases, and underspecified rules that are difficult to resolve through theory alone. Moreover, news corpora evolve over time and differ across cultures, necessitating that researchers revisit the theoretical frameworks underlying these codebooks. In this article, we propose a workflow that uses Large Language Models (LLMs) to augment the creation and refinement of framing codebooks by combining theoretical frameworks with data-driven exploration. Rather than treating LLMs as automated classifiers, this approach positions them as analytic collaborators that help externalize decision rules, surface latent dimensions, and support iterative revisions of codebooks through dialogues between researchers and their data. We illustrate this workflow using a dataset of Latin American news coverage, demonstrating how the application of LLMs' capabilities has led to the surfacing of latent patterns, the generation of frame distinctions, and the adaptation of frameworks to new contexts. This method provides an LLM-assisted strategy that supports methodology creativity while preserving researchers' interpretative authority.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The paper claims to propose a workflow that uses large language models (LLMs) as analytic collaborators rather than automated classifiers to augment the creation and refinement of framing codebooks in deductive content analysis. The workflow combines theoretical frameworks with data-driven exploration through iterative researcher-LLM dialogues that externalize decision rules, surface latent dimensions, and support codebook revisions. It is illustrated with a Latin American news corpus, where the authors report that LLM use surfaced latent patterns, generated frame distinctions, and adapted frameworks to new contexts while preserving researchers' interpretive authority.

Significance. If empirically validated, the proposed workflow would offer a meaningful advance for framing research by providing a structured, human-centered method to resolve ambiguities in codebooks when applied to large, evolving, or cross-cultural news corpora. It correctly positions LLMs as dialogue partners to enhance methodological creativity without ceding control, addressing a practical gap where theory alone proves insufficient. The emphasis on researcher authority is a clear strength. However, the current manuscript presents only a conceptual sketch and qualitative illustration, so its significance is prospective rather than demonstrated.

major comments (1)
  1. [Illustration section] Illustration section (Latin American news application): The demonstration reports qualitative outcomes such as surfaced patterns and adapted frameworks but supplies no evaluation metrics, validation against human coders, inter-rater reliability comparisons, error analysis, or discussion of LLM limitations (e.g., hallucinations or biases). This is load-bearing for the central claim that the workflow reliably supports valid iterative revisions without introducing distortions.
minor comments (1)
  1. [Abstract] Abstract: The phrasing 'the application of LLMs' capabilities has led to...' is vague; specifying the exact dialogue prompts or LLM roles used in the illustration would improve clarity and set appropriate expectations.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive and insightful review. We agree that the illustration section would benefit from greater attention to evaluation and limitations, and we have revised the manuscript to incorporate a more explicit discussion of these issues while maintaining the paper's focus as a methodological proposal rather than an empirical validation study.

read point-by-point responses

  1. Referee: [Illustration section] Illustration section (Latin American news application): The demonstration reports qualitative outcomes such as surfaced patterns and adapted frameworks but supplies no evaluation metrics, validation against human coders, inter-rater reliability comparisons, error analysis, or discussion of LLM limitations (e.g., hallucinations or biases). This is load-bearing for the central claim that the workflow reliably supports valid iterative revisions without introducing distortions.

    Authors: We accept this critique. The illustration is qualitative and designed to show how the workflow operates in practice rather than to empirically prove its reliability across cases. In the revised manuscript we have added a dedicated subsection within the illustration that explicitly discusses LLM limitations (hallucinations, biases, and context drift) and the safeguards the workflow employs through ongoing researcher oversight and iterative prompting. We have also inserted a new Limitations section that outlines the absence of quantitative metrics in the current work and proposes concrete directions for future validation, including comparisons with human coders, inter-rater reliability checks, and error analysis. These additions directly address the concern without converting the paper into an empirical validation study, which would exceed its stated scope as a conceptual workflow contribution. revision: partial

Circularity Check

0 steps flagged

No significant circularity: methodological workflow proposal

full rationale

The paper advances a conceptual workflow for LLM-assisted refinement of framing codebooks in deductive content analysis, illustrated on one news corpus. It contains no equations, parameter fitting, statistical predictions, or first-principles derivations. The central claim is a design sketch that explicitly retains final interpretive authority with human researchers and does not claim automation or error-free output. No load-bearing premise reduces to a self-citation chain, fitted input renamed as prediction, or ansatz smuggled via prior work by the same authors. The argument is self-contained against external benchmarks and does not exhibit any of the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the premise that LLMs possess sufficient capabilities to act as reliable analytic collaborators in surfacing patterns and refining rules while preserving researcher authority; this assumption is not independently evidenced in the abstract.

axioms (1)
  • domain assumption Large language models can effectively externalize decision rules, surface latent dimensions, and support iterative codebook revisions through researcher-model dialogues without compromising validity.
    Core premise invoked throughout the workflow description in the abstract.

pith-pipeline@v0.9.0 · 5525 in / 1295 out tokens · 42933 ms · 2026-05-10T02:30:10.890820+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

58 extracted references · 58 canonical work pages

  1. [1]

    Machine Learning for Social Science: An Agnostic Approach

    Grimmer, Justin and Roberts, Margaret E and Stewart, Brandon M. Machine Learning for Social Science: An Agnostic Approach. Annu. Rev. Polit. Sci. (Palo Alto)

    work page

  2. [2]

    the future of coding

    Than, Nga and Fan, Leanne and Law, Tina and Nelson, Laura K and McCall, Leslie. Updating “the future of coding”: Qualitative coding with generative large language models. Sociol. Methods Res

    work page

  3. [3]

    MetricMate : An interactive tool for generating evaluation criteria for LLM -as-a-judge workflow

    Gebreegziabher, Simret Araya and Chiang, Charles and Wang, Zichu and Ashktorab, Zahra and Brachman, Michelle and Geyer, Werner and Li, Toby Jia-Jun and Gómez-Zará, Diego. MetricMate : An interactive tool for generating evaluation criteria for LLM -as-a-judge workflow. Proceedings of the 4th Annual Symposium on Human-Computer Interaction for Work

    work page

  4. [4]

    Science in the age of large language models

    Birhane, Abeba and Kasirzadeh, Atoosa and Leslie, David and Wachter, Sandra. Science in the age of large language models. Nat. Rev. Phys

    work page

  5. [5]

    Automating democracy: Generative AI , journalism, and the future of democracy

    Arguedas, Amy Ross and Simon, Felix M. Automating democracy: Generative AI , journalism, and the future of democracy

    work page

  6. [6]

    ResearchAgent : Iterative research idea generation over scientific literature with Large Language Models

    Baek, Jinheon and Jauhar, Sujay Kumar and Cucerzan, Silviu and Hwang, Sung Ju. ResearchAgent : Iterative research idea generation over scientific literature with Large Language Models. arXiv [cs.CL]

    work page

  7. [7]

    Title and abstract screening for literature reviews using large language models: an exploratory study in the biomedical domain

    Dennstädt, Fabio and Zink, Johannes and Putora, Paul Martin and Hastings, Janna and Cihoric, Nikola. Title and abstract screening for literature reviews using large language models: an exploratory study in the biomedical domain. Syst. Rev

    work page

  8. [8]

    The content analysis guidebook

    Neuendorf, Kimberly A. The content analysis guidebook

    work page

  9. [9]

    From manual to machine: assessing the efficacy of large language models in content analysis

    Pilny, Andrew and McAninch, Kelly and Slone, Amanda and Moore, Kelsey. From manual to machine: assessing the efficacy of large language models in content analysis. Commun. Res. Rep

    work page

  10. [10]

    Leveraging large language models for literature review tasks - A case study using ChatGPT

    Zimmermann, Robert and Staab, Marina and Nasseri, Mehran and Brandtner, Patrick. Leveraging large language models for literature review tasks - A case study using ChatGPT. Communications in Computer and Information Science

    work page

  11. [11]

    (what) can journalism studies learn from supervised machine learning?

    De Grove, Frederik and Boghe, Kristof and De Marez, Lieven. (what) can journalism studies learn from supervised machine learning?. Journal. Stud

    work page

  12. [12]

    Artificial Intelligence and Journalism

    Broussard, M and Diakopoulos, N and others. Artificial Intelligence and Journalism. Journal. Mass Commun. Q

    work page

  13. [13]

    FACT - GPT : Fact-Checking Augmentation via Claim Matching with LLMs

    Choi, Eun Cheol and Ferrara, Emilio. FACT - GPT : Fact-Checking Augmentation via Claim Matching with LLMs. arXiv [cs.CL]

    work page

  14. [14]

    Hypothesis generation with large language models

    Zhou, Yangqiaoyu and Liu, Haokun and Srivastava, Tejes and Mei, Hongyuan and Tan, Chenhao. Hypothesis generation with large language models. Proceedings of the 1st Workshop on NLP for Science (NLP4Science)

    work page

  15. [15]

    Thematic content analysis using supervised machine learning: An empirical evaluation using German online news

    Scharkow, Michael. Thematic content analysis using supervised machine learning: An empirical evaluation using German online news. Qual. Quant

    work page

  16. [16]

    Big social data analytics in journalism and mass communication: Comparing dictionary-based text analysis and unsupervised topic modeling

    Guo, Lei and Vargo, Chris J and Pan, Zixuan and Ding, Weicong and Ishwar, Prakash. Big social data analytics in journalism and mass communication: Comparing dictionary-based text analysis and unsupervised topic modeling. Journal. Mass Commun. Q

    work page

  17. [17]

    The Use of Supervised Learning Algorithms in Political Communication and Media Studies: Locating Frames in the Press

    García-Marín, Javier and Calatrava, Adolfo. The Use of Supervised Learning Algorithms in Political Communication and Media Studies: Locating Frames in the Press. Communication & Society; Pamplona

    work page

  18. [18]

    ChatGPT for good? On opportunities and challenges of large language models for education

    Kasneci, Enkelejda and Sessler, Kathrin and Küchemann, Stefan and Bannert, Maria and Dementieva, Daryna and Fischer, Frank and Gasser, Urs and Groh, Georg and Günnemann, Stephan and Hüllermeier, Eyke and Krusche, Stephan and Kutyniok, Gitta and Michaeli, Tilman and Nerdel, Claudia and Pfeffer, Jürgen and Poquet, Oleksandra and Sailer, Michael and Schmidt,...

    work page

  19. [19]

    Taking Stock of the Toolkit

    Boumans, Jelle W and Trilling, Damian. Taking Stock of the Toolkit. Digital Journalism

    work page

  20. [20]

    A Survey on Evaluation of Large Language Models

    Chang, Yupeng and Wang, Xu and Wang, Jindong and Wu, Yuan and Yang, Linyi and Zhu, Kaijie and Chen, Hao and Yi, Xiaoyuan and Wang, Cunxiang and Wang, Yidong and Ye, Wei and Zhang, Yue and Chang, Yi and Yu, Philip S and Yang, Qiang and Xie, Xing. A Survey on Evaluation of Large Language Models. ACM Trans. Intell. Syst. Technol

    work page

  21. [21]

    Language models are few-shot learners

    Brown, Tom and Mann, Benjamin and Ryder, Nick and Subbiah, Melanie and Kaplan, Jared D and Dhariwal, Prafulla and Neelakantan, Arvind and Shyam, Pranav and Sastry, Girish and Askell, Amanda and Others. Language models are few-shot learners. Adv. Neural Inf. Process. Syst

    work page

  22. [22]

    Beyond counting words

    Kroon, Anne C and van der Meer, Toni and Vliegenthart, Rens. Beyond counting words. Computational Communication Research

    work page

  23. [23]

    ChatGPT outperforms crowd workers for text-annotation tasks

    Gilardi, Fabrizio and Alizadeh, Meysam and Kubli, Maël. ChatGPT outperforms crowd workers for text-annotation tasks. Proc. Natl. Acad. Sci. U. S. A

    work page

  24. [24]

    Strategy Framing in News Coverage and Electoral Success: An Analysis of Topic Model Networks Approach

    Walter, Dror and Ophir, Yotam. Strategy Framing in News Coverage and Electoral Success: An Analysis of Topic Model Networks Approach. Polit. Commun

    work page

  25. [25]

    Understanding Practices around Computational News Discovery Tools in the Domain of Science Journalism

    Nishal, Sachita and Sinchai, Jasmine and Diakopoulos, Nicholas. Understanding Practices around Computational News Discovery Tools in the Domain of Science Journalism. arXiv [cs.HC]

    work page

  26. [26]

    Navigating the research landscape of algorithm-driven journalism: A systematic literature review of authorship, research trends, and future research pathways

    Sarısakaloğlu, Aynur. Navigating the research landscape of algorithm-driven journalism: A systematic literature review of authorship, research trends, and future research pathways. Journal. Stud

    work page

  27. [27]

    Escape Me If You Can: How AI Reshapes News Organisations’ Dependency on Platform Companies

    Simon, Felix M. Escape Me If You Can: How AI Reshapes News Organisations’ Dependency on Platform Companies. Digital Journalism

    work page

  28. [28]

    Computational social science: Obstacles and opportunities

    Lazer, David and Pentland, Alex and Watts, Duncan J and Aral, Sinan and Athey, Susan and Contractor, Noshir and Freelon, Deen and Gonzalez-Bailon, Sandra and King, Gary and Margetts, Helen and Nelson, Alondra and Salganik, Matthew J and Strohmaier, Markus and Vespignani, Alessandro and Wagner, Claudia. Computational social science: Obstacles and opportuni...

    work page

  29. [29]

    The validity of sentiment analysis: Comparing manual annotation, crowd-coding, dictionary approaches, and machine learning algorithms

    van Atteveldt, Wouter and van der Velden, Mariken A C G and Boukes, Mark. The validity of sentiment analysis: Comparing manual annotation, crowd-coding, dictionary approaches, and machine learning algorithms. Commun. Methods Meas

    work page

  30. [30]

    Recent Advances in Natural Language Processing via Large Pre-trained Language Models: A Survey

    Min, Bonan and Ross, Hayley and Sulem, Elior and Veyseh, Amir Pouran Ben and Nguyen, Thien Huu and Sainz, Oscar and Agirre, Eneko and Heintz, Ilana and Roth, Dan. Recent Advances in Natural Language Processing via Large Pre-trained Language Models: A Survey. ACM Comput. Surv

    work page

  31. [31]

    Deep reinforcement learning from human preferences

    Christiano, P and Leike, J and Brown, Tom B and Martic, Miljan and Legg, S and Amodei, Dario. Deep reinforcement learning from human preferences. Adv. Neural Inf. Process. Syst

    work page

  32. [32]

    Coding latent concepts: a human and LLM -coordinated content analysis procedure

    Fan, Jia and Ai, Yushi and Liu, Xiaofan and Deng, Yilin and Li, Yongning. Coding latent concepts: a human and LLM -coordinated content analysis procedure. Commun. Res. Rep

    work page

  33. [33]

    The media framing dataset: Analyzing news narratives in Mexico and Colombia

    Cuadrado, Juan and Martinez, Elizabeth and Martinez-Santos, Juan Carlos and Puertas, Edwin. The media framing dataset: Analyzing news narratives in Mexico and Colombia. Data Brief

    work page

  34. [34]

    Content analysis

    Krippendorff, Klaus. Content analysis

    work page

  35. [35]

    LLM -assisted content analysis: Using large language models to support deductive coding

    Chew, Robert and Bollenbacher, John and Wenger, Michael and Speer, Jessica and Kim, Annice. LLM -assisted content analysis: Using large language models to support deductive coding. arXiv [cs.CL]

    work page

  36. [36]

    AngleKindling : Supporting Journalistic Angle Ideation with Large Language Models

    Petridis, Savvas and Diakopoulos, Nicholas and Crowston, Kevin and Hansen, Mark and Henderson, Keren and Jastrzebski, Stan and Nickerson, Jeffrey V and Chilton, Lydia B. AngleKindling : Supporting Journalistic Angle Ideation with Large Language Models. Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems

    work page 2023

  37. [37]

    PaTAT : Human- AI collaborative qualitative coding with explainable interactive rule synthesis

    Gebreegziabher, Simret Araya and Zhang, Zheng and Tang, Xiaohang and Meng, Yihao and Glassman, Elena L and Li, Toby Jia-Jun. PaTAT : Human- AI collaborative qualitative coding with explainable interactive rule synthesis. Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems

    work page 2023

  38. [38]

    Do we want less automation?

    Agrawal, Ajay and Gans, Joshua S and Goldfarb, Avi. Do we want less automation?. Science

    work page

  39. [39]

    Bias and unfairness in information retrieval systems: New challenges in the LLM era

    Dai, Sunhao and Xu, Chen and Xu, Shicheng and Pang, Liang and Dong, Zhenhua and Xu, Jun. Bias and unfairness in information retrieval systems: New challenges in the LLM era. Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining

    work page

  40. [40]

    CollabCoder : A lower-barrier, rigorous workflow for inductive collaborative qualitative analysis with large language models

    Gao, Jie and Guo, Yuchen and Lim, Gionnieve and Zhang, Tianqin and Zhang, Zheng and Li, Toby Jia-Jun and Perrault, Simon Tangi. CollabCoder : A lower-barrier, rigorous workflow for inductive collaborative qualitative analysis with large language models. Proceedings of the CHI Conference on Human Factors in Computing Systems

    work page

  41. [41]

    Using Supervised Machine Learning to Code Policy Issues: Can Classifiers Generalize across Contexts?

    Burscher, Bjorn and Vliegenthart, Rens and De Vreese, Claes H. Using Supervised Machine Learning to Code Policy Issues: Can Classifiers Generalize across Contexts?. Ann. Am. Acad. Pol. Soc. Sci

    work page

  42. [42]

    Is anyone responsible?

    Iyengar, Shanto. Is anyone responsible?

    work page

  43. [43]

    Teaching the Computer to Code Frames in News: Comparing Two Supervised Machine Learning Approaches to Frame Analysis

    Burscher, Björn and Odijk, Daan and Vliegenthart, Rens and de Rijke, Maarten and de Vreese, Claes H. Teaching the Computer to Code Frames in News: Comparing Two Supervised Machine Learning Approaches to Frame Analysis. Commun. Methods Meas

    work page

  44. [44]

    News Frame Analysis: An Inductive Mixed-method Computational Approach

    Walter, Dror and Ophir, Yotam. News Frame Analysis: An Inductive Mixed-method Computational Approach. Commun. Methods Meas

    work page

  45. [45]

    Framing European politics: A content analysis of press and television news

    Semetko, Holli A and Valkenburg, Patti M Valkenburg. Framing European politics: A content analysis of press and television news. J. Commun

    work page

  46. [46]

    Pre-Trained Language Models and Their Applications

    Wang, Haifeng and Li, Jiwei and Wu, Hua and Hovy, Eduard and Sun, Yu. Pre-Trained Language Models and Their Applications. Proc. Est. Acad. Sci. Eng

    work page

  47. [47]

    Sparks: Inspiration for Science Writing using Language Models

    Gero, Katy Ilonka and Liu, Vivian and Chilton, Lydia. Sparks: Inspiration for Science Writing using Language Models. Designing Interactive Systems Conference

    work page

  48. [48]

    Three Gaps in Computational Text Analysis Methods for Social Sciences: A Research Agenda

    Baden, Christian and Pipal, Christian and Schoonvelde, Martijn and van der Velden, Mariken A C G. Three Gaps in Computational Text Analysis Methods for Social Sciences: A Research Agenda. Commun. Methods Meas

    work page

  49. [49]

    Computational social science and sociology

    Edelmann, Achim and Wolff, Tom and Montagne, Danielle and Bail, Christopher A. Computational social science and sociology. Annu. Rev. Sociol

    work page

  50. [50]

    The future landscape of large language models in medicine

    Clusmann, Jan and Kolbinger, Fiona R and Muti, Hannah Sophie and Carrero, Zunamys I and Eckardt, Jan-Niklas and Laleh, Narmin Ghaffari and Löffler, Chiara Maria Lavinia and Schwarzkopf, Sophie-Caroline and Unger, Michaela and Veldhuizen, Gregory P and Wagner, Sophia J and Kather, Jakob Nikolas. The future landscape of large language models in medicine. Co...

    work page

  51. [51]

    Guilty by association: Using word embeddings to measure ethnic stereotypes in news coverage

    Kroon, Anne C and Trilling, Damian and Raats, Tamara. Guilty by association: Using word embeddings to measure ethnic stereotypes in news coverage. Journal. Mass Commun. Q

    work page

  52. [52]

    Proposing an Open-Sourced Tool for Computational Framing Analysis of Multilingual Data

    Guo, Lei and Su, Chao and Paik, Sejin and Bhatia, Vibhu and Akavoor, Vidya Prasad and Gao, Ge and Betke, Margrit and Wijaya, Derry. Proposing an Open-Sourced Tool for Computational Framing Analysis of Multilingual Data. Digital Journalism

    work page

  53. [53]

    Envisioning the Applications and Implications of Generative AI for News Media

    Nishal, Sachita and Diakopoulos, Nicholas. Envisioning the Applications and Implications of Generative AI for News Media. arXiv [cs.CY]

    work page

  54. [54]

    Why we support and encourage the use of large language models in NEJM AI submissions

    Koller, Daphne and Beam, Andrew and Manrai, Arjun and Ashley, Euan and Liu, Xiaoxuan and Gichoya, Judy and Holmes, Chris and Zou, James and Dagan, Noa and Wong, Tien Y and Blumenthal, David and Kohane, Isaac. Why we support and encourage the use of large language models in NEJM AI submissions. NEJM AI

    work page

  55. [55]

    [Data set for replication] Valenzuela, S ., Piña, M ., & Ramírez, J

    Valenzuela, Sebastián. [Data set for replication] Valenzuela, S ., Piña, M ., & Ramírez, J . (2017). Behavioral effects of framing on social media users: How conflict, economic, human interest, and morality frames drive news sharing. Journal of Communication, 67(5), 803-826

    work page 2017

  56. [56]

    Supporting Qualitative Analysis with Large Language Models: Combining Codebook with GPT -3 for Deductive Coding

    Xiao, Ziang and Yuan, Xingdi and Liao, Q Vera and Abdelghani, Rania and Oudeyer, Pierre-Yves. Supporting Qualitative Analysis with Large Language Models: Combining Codebook with GPT -3 for Deductive Coding. Companion Proceedings of the 28th International Conference on Intelligent User Interfaces

    work page

  57. [57]

    conversing

    Hayes, Adam S. “conversing” with qualitative data: Enhancing qualitative research through large Language Models ( LLMs ). Int. J. Qual. Methods

    work page

  58. [58]

    A Computational Inflection for Scientific Discovery

    Hope, Tom and Downey, Doug and Weld, Daniel S and Etzioni, Oren and Horvitz, Eric and Shein, Esther. A Computational Inflection for Scientific Discovery. Communications of the ACM

    work page