Pith. sign in

REVIEW 3 major objections 5 minor 43 references

Refined and Segmented Price Sentiment Indices from Survey Comments

T0 review · 3 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper constructs price sentiment indices from Japanese survey comments using LLM classification and segmentation, and claims higher correlations with CPI, CGPI, and SPPI than the prior word-count index.

desk verdict Solid LLM-based price sentiment index extension; the claimed win over the BoJ baseline is not yet established because the baseline row may come from a different sample and seasonal-adjustment scheme. read the letter →

arxiv 2411.09937 v2 pith:YGP45ET3 submitted 2024-11-15 cs.CL q-fin.CP

Masahiro Suzuki , Hiroki Sakaji This is my paper · ORCID
classification cs.CLq-fin.CP
keywords pricesentimentindexEconomyWatchersSurveylargelanguagemodelsinflationforecastingconsumerproducertextclassificationJapan
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to build better price sentiment indicators for Japan by reading the monthly Economy Watchers Survey comments with large language models instead of counting words. It claims that an index assembled from LLM-classified comments correlates more strongly with Japan's consumer and corporate price indices than the word-count index from earlier work. Because the survey carries metadata on whether a comment comes from a household or a business and from a goods or services industry, the authors can split the comments into five specialized indices, and they show the consumer-focused splits track the corresponding consumer price components better than the aggregate index. Getting this right matters because a sentiment measure that leads official price statistics by several months could support earlier inflation signals for policy and investment decisions.

What carries the argument

The carrying mechanism is the four-step pipeline: (1) FinBERT filters comments about prices; (2) several LLMs classify each comment as rising, stable, falling, or not price-related, each returning a confidence and a reason; (3) a separate LLM synthesizes these outputs into one label; (4) comments are aggregated by month and segment into PSI = (Rise - Fall) / (Rise + Fall + Stable). Segmentation uses the survey's comment domain (household vs corporate trends) and a manual mapping of 169 respondent industries into manufacturing or non-manufacturing, yielding five specific indices plus a general one. The formula is the same one used by earlier work, so the claimed performance gain comes from the classification and segmentation rather than from a new aggregation formula.

What would settle it

Recompute the [7] word-count price sentiment index on the identical January 2001 to June 2024 sample and run the same lagged-correlation search; if the recomputed baseline matches the column shown in Table VI, then re-test the comparison. Alternatively, hold out 2025 onward data and check whether the LLM-based index still out-correlates the baseline out-of-sample.

Watch

Extended reading notes

Core claim

The central claim is that a price sentiment index (PSI) built from LLM-classified survey comments outperforms the prior word-based PSI for every official price index studied: the general PSI attains higher maximum lagged correlations than baseline [7] for core-core CPI, CPI goods, CPI services, CGPI, and SPPI (e.g., 0.635 vs 0.583 for core-core CPI at a 14-month lead), and the segmented consumer PSIs improve further on the three consumer indices. The paper also reports Granger causality from the general PSI to core-core CPI, CGPI, and SPPI at the 1% level with 12-month lags, which supports the interpretation that the index leads rather than merely tracks official prices. The improvement is attributed to two changes: replacing word counting with LLM classification (with fine-tuned FinBERT filtering price-related comments, and an ensemble of GPT-4o, Claude 3.5 Sonnet, and Gemini 1.5 Flash whose outputs are merged by another LLM), and segmenting comments by the survey's domain and respondent industry before aggregation.

Load-bearing premise

The load-bearing premise is that the baseline index from the prior study was measured on the same January 2001 to June 2024 sample and with the same lag search; if the prior study used different dates, the reported improvements could come from sample differences rather than the new method.

Editorial extensions

If this is right

  • For the three consumer price indices (core-core CPI, goods, services), the segmented consumer PSIs show higher time-lagged correlations than both the general PSI and the baseline, so tracking household-only comments helps nowcast consumer prices.
  • The general PSI Granger-causes core-core CPI, CGPI, and SPPI at 1% significance with 12-month lags, implying the index contains predictive content beyond contemporaneous correlation.
  • Corporate-goods and corporate-services PSIs have small sample sizes (about 25 comments per month), so the general PSI remains the better vehicle for tracking CGPI and SPPI.
  • Because classification uses 5-shot in-context learning with API models, the same pipeline can be applied to other monthly surveys without retraining.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A cheaper variant using a single strong classifier or majority voting across model outputs might achieve most of the integration gain; the paper only tests LLM-based integration, so a simple aggregation baseline would clarify how much the integrator adds.
  • The manual mapping of 169 respondent industries to manufacturing/non-manufacturing is a hidden judgment call; an automated or sensitivity-checked mapping would show whether segmentation results are robust to that choice.
  • The method implicitly assumes that comments mentioning prices are representative of actual price movements; extending the index to subnational or sectoral price indices would test whether segmentation generalizes beyond national aggregates.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes a pipeline for constructing price sentiment indices (PSIs) from the Cabinet Office's Economy Watchers Survey. Price-related comments are filtered with a fine-tuned FinBERT model; the direction of price movements (rise/stable/fall) is classified by multiple LLMs (GPT-4o, Claude 3.5 Sonnet, Gemini 1.5 Flash) with 5-shot in-context learning, and their outputs are integrated by an LLM. The classified comments are aggregated monthly into a General PSI and five segmented PSIs (consumer general, consumer goods, consumer services, corporate goods, corporate services) by using the survey domain and respondent industry. The indices are compared with core-core CPI, CPI Goods, CPI Services, CGPI, and SPPI via maximum time-lagged correlations, and Granger causality tests are reported. The main claim is that the proposed PSIs have higher correlations than the previous word-count-based index of Nakajima et al. [7].

Significance. The paper has several strengths: the classification tasks are evaluated on labeled data with several models; the data version (2024.07.0) and API model versions are specified; the seasonal-adjustment sensitivity is examined in Appendix D; and the Granger causality tests provide a direct test of predictive content. If the comparative claim against [7] were established on a common sample, the result would be a useful demonstration that LLM-based, segmented sentiment indices add information for tracking Japanese price trends. However, as presented, the headline comparison is not yet a controlled comparison, and the statistical significance of the correlation improvements is not quantified. The contribution is therefore conditionally useful: the methodological pieces are sound, but the central evaluation needs strengthening.

major comments (3)
  1. [Section IV, Table VI and Appendix D] The central comparative claim that the proposed General PSI and Specific PSIs achieve higher time-lagged correlations than the baseline [7] is not established because the baseline values in Table VI are not shown to be computed on the same sample and index definitions. Appendix D states that [7] uses seasonally adjusted CPI, whereas this paper uses non-seasonally adjusted indices, and the manuscript does not state that the [7] index was recomputed on the January 2001 to June 2024 sample described in Section III-C. Since [7] is a 2021 Bank of Japan paper, its reported correlations may come from a different sample period and possibly a different survey data version. The differences are small in some cases (e.g., 0.635 vs 0.583 for core-core CPI, 0.793 vs 0.778 for CGPI), so they could be affected by sample-period, seasonal-adjustment, or index-revision differences. Please either recompute the baseline on the same sample with identical preprocessing and lag search, or clearly qualify the comparison and present the original [7] values as references rather than as a controlled benchmark.
  2. [Section IV, Table VI] The reported correlations are maxima over a lag grid (the values in parentheses are the lags at which the maximum is attained), and the paper gives no confidence intervals or significance tests for these correlations or for the differences between the proposed indices and the baseline. Under the null that two series are independent, the distribution of the maximum lagged correlation over a grid of candidate lags stochastically dominates that of a fixed-lag correlation, so the point estimates are likely upward-biased. In addition, the comparison across five target indices and multiple PSI variants involves multiple testing. The authors should report the full lag-correlation curves or, at minimum, provide standard errors and a test of whether the improvement over the baseline is significant, and discuss the selection of the lag and the integration model as part of the procedure.
  3. [Section III-B and Table IV] The selection of Gemini 1.5 Flash as the integration model (Table IV) was based on the test-set performance of the integration step, and the same test-set evaluations informed the choice of the constituent models. This selection is an additional data-dependent degree of freedom that is not accounted for when the resulting PSI is later evaluated for correlation with price indices. The paper should either use a nested or separate validation split for the integration-model choice, or report sensitivity of Table VI to alternative integration models (e.g., Claude 3.5 Sonnet), so that the reported correlation improvements cannot be attributed to overfitting the integration choice to the test set.
minor comments (5)
  1. [Section III-A] There is a typo: 'fune-tuned' should be 'fine-tuned'; also in Table I, 'T HE' should be 'THE', and in the NOTES section, 'the their affiliated institutions' should be 'their affiliated institutions'.
  2. [Section III-A and III-B] The paper does not report inter-annotator agreement for the manual labeling of the 304 comments in Price Direction Data 1 or for the price-direction labels of the 1,000 comments in Price Direction Data 2; since these labels form the gold standard for the classification evaluation, a measure of agreement (e.g., Cohen's kappa) would strengthen the results.
  3. [Section IV, Table VIII] The Corporate Goods PSI and Corporate Services PSI are constructed from only about 25 comments per month; the paper discusses this limitation qualitatively, but it would be useful to show the time-series volatility or confidence bands for these indices to help readers gauge their reliability.
  4. [Section IV, paragraph on lag correlations] The sentence 'When examining the lag correlation between CPI and SPPI, the maximum value of 0.920 is observed when CPI leads by three months' reports a correlation between two existing price indices, not between a PSI and a price index; clarify that this is a reference point from the data rather than a result about the proposed indices.
  5. [Equation (1), Section III-C] The definition PSI = (Rise - Fall)/(Rise + Fall + Stable) is stated, but the paper does not specify whether the baseline [7] used the same normalization; a sentence describing the baseline construction would remove ambiguity about the comparability of levels.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the PSIs are built from survey comments with manual labels and validated against external price indices; the baseline-comparison concern is sample comparability, not a definitional reduction.

full rationale

The derivation chain is self-contained against external benchmarks. Price-related comments are filtered by FinBERT trained on 1,000 manually labeled survey comments, and price directions are classified by LLMs evaluated on 1,304 manually labeled comments; neither label set is derived from CPI, CGPI, or SPPI values. The PSI aggregation in Eq. (1) is a fixed ratio formula adopted from prior work [7] and contains no parameter fitted to the price indices, so the correlations in Table VI are external validation rather than a restatement of the inputs. The maximum-lag correlations and Granger causality tests are descriptive and inferential assessments of the constructed index against official statistics, not fitted predictions equivalent to their targets. The only substantive concern is that the Baseline [7] row in Table VI is not shown to be recomputed on the same January 2001 to June 2024 sample, and Appendix D concedes that [7] used seasonally adjusted CPI while this paper uses non-seasonally adjusted indices; this is a benchmark-comparability flaw affecting the strength of the headline claim, not a circularity reduction. Self-citations, such as [35] for the dataset and [36]/[37] for FinBERT, supply tools and data but are not load-bearing arguments that define the target result, and no uniqueness theorem or ansatz is smuggled in through citations. No equation reduces to its own input and no fitted parameter is renamed as a prediction, so the circularity score is 0.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The index construction itself has no fitted constants, but three hand-chosen or post hoc choices carry weight: the lag at which correlations are maximized, the choice of Gemini 1.5 Flash as final integrator after seeing test F1, and the manual mapping of 169 industries into manufacturing/non-manufacturing. The main domain assumptions are that survey comments are informative about price trends, that the survey's domain and industry labels correspond to the economic segments, and that the manually annotated labels are reliable ground truth. No new physical or economic entities are introduced.

free parameters (3)
  • lag_at_max_correlation = 14 (core core CPI), 7 (goods), 16 (services), 3 (CGPI), 15 (SPPI)
    The headline correlations are maxima over a lag grid; the lag is chosen post hoc, making the reported coefficient a fitted statistic.
  • integration_model_selection = Gemini 1.5 Flash
    The final integrator was chosen after observing test-set weighted F1 (88.0 vs 85.3 for the best single model), and cost was then used as justification.
  • industry_manufacturing_mapping = 169 industry categories manually assigned
    The respondent industries are manually classified into manufacturing and non-manufacturing after parentheses are removed; this mapping is hand-chosen and underlies all segmented PSIs.
assumptions (4)
  • domain assumption Economy Watchers Survey comments are informative about actual price trends.
    All indices are built from these comments; if comments are uninformative, correlations with official indices would be spurious. Invoked in Section III-A.
  • domain assumption Comment domain and respondent industry map cleanly to consumer/corporate and goods/services segments.
    Used to build the five segmented PSIs in Section III-C and Table V; overlap between household and corporate perspectives may exist.
  • standard math Maximum lagged Pearson correlation and Granger causality at 12 lags are valid measures of leading information.
    Used to claim leading correlations in Section IV, Tables VI and VII, with no correction for multiple lag comparisons.
  • domain assumption Manual labels by the authors are ground truth for filtering and direction classification.
    All classification F1 scores and the subsequent index construction depend on these labels; no inter-annotator agreement is reported. See Section III-A and III-B.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Refined and Segmented Price Sentiment Indices from Survey Comments." pith.science (2026). https://pith.science/paper/YGP45ET3

@misc{pith2026241109937,
  author       = {Pith},
  title        = {Pith review of: Refined and Segmented Price Sentiment Indices from Survey Comments},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YGP45ET3}},
  note         = {Machine review of arXiv:2411.09937}
}
read the original abstract

We aim to enhance a price sentiment index and to more precisely understand price trends from the perspective of not only consumers but also businesses. We extract comments related to prices from the Economy Watchers Survey conducted by the Cabinet Office of Japan and classify price trends using a large language model (LLM). We classify whether the survey sample reflects the perspective of consumers or businesses, and whether the comments pertain to goods or services by utilizing information on the fields of comments and the industries of respondents included in the Economy Watchers Survey. From these classified price-related comments, we construct price sentiment indices not only for a general purpose but also for more specific objectives by combining perspectives on consumers and prices, as well as goods and services. It becomes possible to achieve a more accurate classification of price directions by employing a LLM for classification. Furthermore, integrating the outputs of multiple LLMs suggests the potential for the better performance of the classification. The use of more accurately classified comments allows for the construction of an index with a higher correlation to existing indices than previous studies. We demonstrate that the correlation of the price index for consumers, which has a larger sample size, is further enhanced by selecting comments for aggregation based on the industry of the survey respondents.

Figures

Figures reproduced from arXiv: 2411.09937 by the authors.

Figure 1
Figure 1. Overview processes of aggregating indicies from the Economy Watchers Survey. (1) Extract comments related to prices from the Economy Watchers [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 39 canonical work pages

  1. [6]

    Economic analysis using machine learning: Text mining of the Economy Watchers Survey,

    K. Otaka and K. Kan, “Economic analysis using machine learning: Text mining of the Economy Watchers Survey,” Bank of Japan Working Paper Series, Tech. Rep., 2018

    work page 2018

  2. [7]

    Extracting firms’ short-term inflation expectations from the Economy Watchers Survey using text analysis,

    J. Nakajima, H. Yamagata, T. Okuda, S. Katsuki, and T. Shinohara, “Extracting firms’ short-term inflation expectations from the Economy Watchers Survey using text analysis,” Bank of Japan, Tech. Rep., 2021

    work page 2021

  3. [25]

    Ex- tracting firms’ short-term inflation expectations from survey comments using text analysis,

    J. Nakajima, T. Okuda, H. Yamagata, S. Katsuki, and T. Shinohara, “Ex- tracting firms’ short-term inflation expectations from survey comments using text analysis,” 2022

    work page 2022

  4. [1]

    Estimation of firms’ inflation expectations using the survey DI,

    J. Nakajima, “Estimation of firms’ inflation expectations using the survey DI,” Institute of Economic Research, Hitotsubashi University, Tech. Rep., 2023

    work page 2023

  5. [2]

    Forecasting CPI inflation components with Hierarchical Recurrent Neural Networks,

    O. Barkan, J. Benchimol, I. Caspi, E. Cohen, A. Hammer, and N. Koenigstein, “Forecasting CPI inflation components with Hierarchical Recurrent Neural Networks,” International Journal of Forecasting, vol. 39, no. 3, pp. 1145–1162, 2023. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0169207022000607

    work page 2023

  6. [3]

    Forecasting UK inflation bottom up,

    A. Joseph, G. Potjagailo, C. Chakraborty, and G. Kapetanios, “Forecasting UK inflation bottom up,” International Journal of Forecasting, vol. 40, no. 4, pp. 1521–1538, 2024. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0169207024000013

    work page 2024

  7. [4]

    Internet search behavior as an economic forecasting tool: The case of inflation expectations,

    G. Guzman, “Internet search behavior as an economic forecasting tool: The case of inflation expectations,” Journal of economic and social measurement, vol. 36, no. 3, pp. 119–167, 2011

    work page 2011

  8. [5]

    Nowcasting prices using Google trends: an application to Central America,

    S. Seabold and A. Coppola, “Nowcasting prices using Google trends: an application to Central America,” World Bank Policy Research Working Paper, no. 7398, 2015

    work page 2015

Show all 43 references
  1. [8]

    Development and anal- ysis of medical instruction-tuning for Japanese large language models,

    I. Sukeda, M. Suzuki, H. Sakaji, and S. Kodera, “Development and anal- ysis of medical instruction-tuning for Japanese large language models,” AIH, vol. 1, no. 2, p. 107, 2024

    2024

  2. [9]

    Large Language Models are legal but they are not: Making the case for a powerful LegalLLM,

    T. Jayakumar, F. Farooqui, and L. Farooqui, “Large Language Models are legal but they are not: Making the case for a powerful LegalLLM,” in Proceedings of the Natural Legal Language Processing Workshop 2023 , Dec. 2023, pp. 223–229. [Online]. Available: https://aclanthology.or...

    2023

  3. [10]

    JaFIn: Japanese Financial Instruction Dataset,

    K. Tanabe, M. Suzuki, H. Sakaji, and I. Noda, “JaFIn: Japanese Financial Instruction Dataset,” 2024. [Online]. Available: https: //arxiv.org/abs/2404.09260

    2024 arXiv

  4. [11]

    Language models are few-shot learners,

    T. Brown, B. Mann, N. Ryder, M. Subbiah, J. D. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askellet al., “Language models are few-shot learners,” in Advances in neural information processing systems, vol. 33, 2020, pp. 1877–1901

    2020

  5. [12]

    Large language models are zero-shot reasoners,

    T. Kojima, S. S. Gu, M. Reid, Y . Matsuo, and Y . Iwasawa, “Large language models are zero-shot reasoners,” in Advances in neural infor- mation processing systems , vol. 35, 2022, pp. 22 199–22 213

    2022

  6. [13]

    The causal relationship between producer price index and consumer price index: Empirical evidence from selected European countries,

    S. Akcay, “The causal relationship between producer price index and consumer price index: Empirical evidence from selected European countries,” International Journal of Economics and Finance , vol. 3, no. 6, pp. 227–232, 2011

    2011

  7. [14]

    Exchange rate pass- through to Japanese prices: Import prices, producer prices, and the core CPI,

    Y . Sasaki, Y . Yoshida, and P. K. Otsubo, “Exchange rate pass- through to Japanese prices: Import prices, producer prices, and the core CPI,” Journal of International Money and Finance , vol. 123, p. 102599, 2022. [Online]. Available: https://www.sciencedirect.com/ science/ar...

    2022

  8. [15]

    Dynamic causality between PPI and CPI in China: A rolling window bootstrap approach,

    J. Sun, J. Xu, X. Cheng, J. Miao, and H. Mu, “Dynamic causality between PPI and CPI in China: A rolling window bootstrap approach,” International Journal of Finance & Economics, vol. 28, no. 2, pp. 1279– 1289, 2023

    2023

  9. [16]

    The consumer price index prediction using machine learning approaches: Evidence from the united states,

    T.-T. Nguyen, H.-G. Nguyen, J.-Y . Lee, Y .-L. Wang, and C.-S. Tsai, “The consumer price index prediction using machine learning approaches: Evidence from the united states,” Heliyon, vol. 9, no. 10, 2023

    2023

  10. [17]

    Comparison four kernels of svr to predict consumer price index,

    M. Rohmah, I. Putra, R. Hartati, and L. Ardiantoro, “Comparison four kernels of svr to predict consumer price index,” in Journal of Physics: Conference Series, vol. 1737, no. 1. IOP Publishing, 2021, p. 012018

    2021

  11. [18]

    Predicting Socio-Economic Indicators using News Events,

    S. Chakraborty, A. Venkataraman, S. Jagabathula, and L. Subramanian, “Predicting Socio-Economic Indicators using News Events,” in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , ser. KDD ’16, 2016, p. 1455–1464. [Online]. Av...

    2016 doi

  12. [19]

    Measuring news sentiment,

    A. H. Shapiro, M. Sudhof, and D. J. Wilson, “Measuring news sentiment,” Journal of Econometrics , vol. 228, no. 2, pp. 221– 243, 2022. [Online]. Available: https://www.sciencedirect.com/science/ article/pii/S0304407620303535

    2022

  13. [20]

    Can we measure inflation expectations using twitter?

    C. Angelico, J. Marcucci, M. Miccoli, and F. Quarta, “Can we measure inflation expectations using twitter?” Journal of Econometrics , vol. 228, no. 2, pp. 259–277, 2022. [Online]. Available: https: //www.sciencedirect.com/science/article/pii/S0304407622000227

    2022

  14. [21]

    Sentiment Summarization of Financial Reports by LSTM RNN model with the Japan Economic Watcher Survey Data,

    Y . Yamamoto and Y . Matsuo, “Sentiment Summarization of Financial Reports by LSTM RNN model with the Japan Economic Watcher Survey Data,” in Proceedings of the Annual Conference of JSAI , 2016, pp. 3L3OS16a2–3L3OS16a2, (in Japanese)

    2016

  15. [22]

    News-based business sentiment and its properties as an economic index,

    K. Seki, Y . Ikuta, and Y . Matsubayashi, “News-based business sentiment and its properties as an economic index,” Information Processing & Management, vol. 59, no. 2, p. 102795, 2022. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0306457321002739

    2022

  16. [23]

    Stock Selection Attempt using Sentiment of Japan Company Handbook,

    M. Suzuki, “Stock Selection Attempt using Sentiment of Japan Company Handbook,” in Proceedings of the Annual Conference of JSAI, 2024, pp. 2I6GS1001–2I6GS1001, (in Japanese)

    2024

  17. [24]

    Forecasting Japanese inflation with a news-based leading indicator of economic activities,

    K. Goshima, H. Ishijima, M. Shintani, and H. Yamamoto, “Forecasting Japanese inflation with a news-based leading indicator of economic activities,” Studies in Nonlinear Dynamics & Econometrics , vol. 25, no. 4, pp. 111–133, 2021

    2021

  18. [26]

    Is ChatGPT a Financial Expert? Evaluating Language Models on Financial Natural Language Processing,

    Y . Guo, Z. Xu, and Y . Yang, “Is ChatGPT a Financial Expert? Evaluating Language Models on Financial Natural Language Processing,” in Findings of the Association for Computational Linguistics: EMNLP 2023 , Dec. 2023, pp. 815–821. [Online]. Available: https: //aclanthology.org...

    2023

  19. [27]

    Are ChatGPT and GPT-4 General-Purpose Solvers for Financial Text Analytics? A Study on Several Typical Tasks,

    X. Li, S. Chan, X. Zhu, Y . Pei, Z. Ma, X. Liu, and S. Shah, “Are ChatGPT and GPT-4 General-Purpose Solvers for Financial Text Analytics? A Study on Several Typical Tasks,” in Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: Industry Trac...

    2023

  20. [28]

    FinBen: A Holistic Financial Benchmark for Large Language Models,

    Q. Xie, W. Han, Z. Chen, R. Xiang, X. Zhang, Y . He, M. Xiao, D. Li, Y . Dai, D. Feng, Y . Xu, H. Kang, Z. Kuang, C. Yuan, K. Yang, Z. Luo, T. Zhang, Z. Liu, G. Xiong, Z. Deng, Y . Jiang, Z. Yao, H. Li, Y . Yu, G. Hu, J. Huang, X.-Y . Liu, A. Lopez-Lira, B. Wang, Y . Lai, H. W...

    2024 arXiv

  21. [29]

    Construction of a Japanese Financial Benchmark for Large Language Models,

    M. Hirano, “Construction of a Japanese Financial Benchmark for Large Language Models,” in Proceedings of the Joint Workshop of the 7th Financial Technology and Natural Language Processing, the 5th Knowledge Discovery from Unstructured Data in Financial Services, and the 4th Wo...

    2024

  22. [30]

    Chatgpt-based investment portfolio selection,

    O. Romanko, A. Narayan, and R. H. Kwon, “Chatgpt-based investment portfolio selection,” in Operations Research Forum , vol. 4, no. 4. Springer, 2023, p. 91

    2023

  23. [31]

    Can ChatGPT improve investment decisions? From a portfolio management perspective,

    H. Ko and J. Lee, “Can ChatGPT improve investment decisions? From a portfolio management perspective,” Finance Research Letters, vol. 64, p. 105433, 2024. [Online]. Available: https://www.sciencedirect.com/ science/article/pii/S154461232400463X

    2024

  24. [32]

    Can ChatGPT assist in picking stocks?

    M. Pelster and J. Val, “Can ChatGPT assist in picking stocks?” Finance Research Letters , vol. 59, p. 104786, 2024. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S1544612323011583

    2024

  25. [33]

    LLMFactor: Extracting Profitable Factors through Prompts for Explainable Stock Movement Prediction,

    M. Wang, K. Izumi, and H. Sakaji, “LLMFactor: Extracting Profitable Factors through Prompts for Explainable Stock Movement Prediction,” in Findings of the Association for Computational Linguistics ACL 2024 , Aug. 2024, pp. 3120–3131. [Online]. Available: https://aclanthology.o...

    2024

  26. [34]

    Learning to Generate Explainable Stock Predictions using Self-Reflective Large Language Models,

    K. J. Koa, Y . Ma, R. Ng, and T.-S. Chua, “Learning to Generate Explainable Stock Predictions using Self-Reflective Large Language Models,” in Proceedings of the ACM Web Conference 2024, ser. WWW ’24, 2024, p. 4304–4315. [Online]. Available: https://doi.org/10.1145/3589334.3645611

    2024

  27. [35]

    Economy Watchers Survey provides Datasets and Tasks for Japanese Financial Domain,

    M. Suzuki and H. Sakaji, “Economy Watchers Survey provides Datasets and Tasks for Japanese Financial Domain,” 2024. [Online]. Available: https://arxiv.org/abs/2407.14727

    2024 arXiv

  28. [36]

    FinDeBERTaV2: Word-Segmentation-Free Pre-trained Language Model for Finance,

    M. Suzuki, H. Sakaji, M. Hirano, and K. Izumi, “FinDeBERTaV2: Word-Segmentation-Free Pre-trained Language Model for Finance,” Transactions of the Japanese Society for Artificial Intelligence , vol. 39, no. 4, pp. FIN23–G 1–14, 2024, (in Japanese)

    2024

  29. [37]

    Constructing and analyzing domain-specific language model for financial text mining,

    ——, “Constructing and analyzing domain-specific language model for financial text mining,” Information Processing & Management , vol. 60, no. 2, p. 103194, 2023

    2023

  30. [38]

    Sudachi: a Japanese Tokenizer for Business,

    K. Takaoka, S. Hisamoto, N. Kawahara, M. Sakamoto, Y . Uchida, and Y . Matsumoto, “Sudachi: a Japanese Tokenizer for Business,” in Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018) , 2018. [Online]. Available: https://aclanth...

    2018

  31. [39]

    From Base to Conversational: Japanese Instruction Dataset and Tuning Large Language Models,

    M. Suzuki, M. Hirano, and H. Sakaji, “From Base to Conversational: Japanese Instruction Dataset and Tuning Large Language Models,” in 2023 IEEE International Conference on Big Data (Big Data) , 2023, pp. 5684–5693

    2023

  32. [40]

    LLM-jp: A Cross-organizational Project for the Research and Development of Fully Open Japanese LLMs,

    LLM-jp, “LLM-jp: A Cross-organizational Project for the Research and Development of Fully Open Japanese LLMs,” 2024. [Online]. Available: https://arxiv.org/abs/2407.03963

    2024 arXiv

  33. [41]

    Continual Pre-Training for Cross-Lingual LLM Adaptation: Enhancing Japanese Language Capa- bilities,

    K. Fujii, T. Nakamura, M. Loem, H. Iida, M. Ohi, K. Hattori, H. Shota, S. Mizuki, R. Yokota, and N. Okazaki, “Continual Pre-Training for Cross-Lingual LLM Adaptation: Enhancing Japanese Language Capa- bilities,” in Proceedings of the First Conference on Language Modeling, ser....

    2024

  34. [42]

    Fine-tuned models: Fine-tuned parameters in FinBERT and DeBERTaV2 models are shown in Table IX

  35. [43]

    Yes”, and if it does not, answer “No

    Prompts for LLM: The following boxes show examples of prompts used for the LLMs in the sections on price-related texts filtration (Section III-A) and price direction classification (Section III-B). We manually develop the confidence level and reasoning used for in-context lear...

    2010

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.