Browse State-of-the-Art › Sentiment Analysis

Sentiment Analysis

1,509 papers with code · 42 benchmarks · 104 datasets archive 2025-07-28

Computer VisionNatural Language Processing

Sentiment Analysis is the task of classifying the polarity of a given text. For instance, a text-based tweet can be categorized into either "positive", "negative", or "neutral". Given the text and accompanying labels, a model can be trained to predict the correct sentiment.

Sentiment Analysis techniques can be categorized into machine learning approaches, lexicon-based approaches, and even hybrid methods. Some subcategories of research in sentiment analysis include: multimodal sentiment analysis, aspect-based sentiment analysis, fine-grained opinion analysis, language specific sentiment analysis.

More recently, deep learning techniques, such as RoBERTa and T5, are used to train high-performing sentiment classifiers that are evaluated using metrics like F1, recall, and precision. To evaluate sentiment analysis systems, benchmark datasets like SST, GLUE, and IMDB movie reviews are used.

Further readings:

  • Sentiment Analysis Based on Deep Learning: A Comparative Study

Description from the archive archive 2025-07-28.

Benchmarks archive 2025-07-28

45 leaderboard tables shown for this task, 42 with rows (a “benchmark” on this site is a table with at least one row, as on /sota), ordered by row count. “Best model” is the first row in the archive's own order at snapshot; nothing is re-ranked here and metric direction is not recorded in the archive. PwC's Trend sparklines are not in the archive, so that column is omitted. 10 shown of 45 until expanded.

Syntology 23 newer papers were judged to report on this task's benchmarks so far (Claude Sonnet 4.5, reading their tables: a model's judgement, not a result; 0 placed): SST-2 Binary classification (17 judged, 0 placed), IMDb (3 judged, 0 placed), SST-5 Fine-grained classification (4 judged, 0 placed), MR (3 judged, 0 placed). Drawn from 8,886 of 18,637 newer papers checked so far (733 more had nothing to place; 41 could not be read).

DatasetBest model (first row in archive order)PaperCodeSyntologyCompare
SST-2 Binary classification (87 rows) T5-11B Exploring the Limits of Transfer Learning with a Unified... code Syntology 21 ran (of which 0 constructed an object rather than computing a result; 20 with no instrument failure: 1 honoured, 0 violated, 19 with no contract checked; 1 where Syntology's instrument failed) · 10 unverified Compare
IMDb (49 rows) RoBERTa-large with LlamBERT LlamBERT: Large-scale low-cost data annotation in NLP code — Compare
SST-5 Fine-grained classification (31 rows) Llama-3.3-70B + CAPO CAPO: Cost-Aware Prompt Optimization code — Compare
Yelp Binary classification (20 rows) XLNet XLNet: Generalized Autoregressive Pretraining for Language Understanding code Syntology official (archive's flag): 1 ran · 15 ran (of which 2 constructed an object rather than computing a result; 10 with no instrument failure: 2 honoured, 0 violated, 8 with no contract checked; 5 where Syntology's instrument failed) · 9 unverified Compare
MR (19 rows) VLAWE Vector of Locally-Aggregated Word Embeddings (VLAWE): A Novel... code — Compare
Yelp Fine-grained classification (17 rows) XLNet XLNet: Generalized Autoregressive Pretraining for Language Understanding code Syntology official (archive's flag): 1 ran · 15 ran (of which 2 constructed an object rather than computing a result; 10 with no instrument failure: 2 honoured, 0 violated, 8 with no contract checked; 5 where Syntology's instrument failed) · 9 unverified Compare
BanglaBook (13 rows) Bangla-BERT (large) BanglaBook: A Large-scale Bangla Dataset for Sentiment Analysis... code — Compare
DynaSent (12 rows) GPT-4o Fine-Tuned (Minimal) ELECTRA and GPT-4o: Cost-Effective Partners for Sentiment Analysis code — Compare
SST-3 (11 rows) GPT-4o-mini Fine-Tuned ELECTRA and GPT-4o: Cost-Effective Partners for Sentiment Analysis code — Compare
Sentiment Merged (10 rows) GPT-4o Fine-Tuned (Minimal) ELECTRA and GPT-4o: Cost-Effective Partners for Sentiment Analysis code — Compare
User and product information (10 rows) MA-BERT MA-BERT: Learning Representation by Incorporating Multi-Attribute... code — Compare
Amazon Review Full (9 rows) BERT large Unsupervised Data Augmentation for Consistency Training code Syntology official (archive's flag): 10 ran · 30 ran (of which 3 constructed an object rather than computing a result; 22 with no instrument failure: 2 honoured, 0 violated, 20 with no contract checked; 8 where Syntology's instrument failed) · 22 unverified Compare
Amazon Review Polarity (9 rows) BERT large Unsupervised Data Augmentation for Consistency Training code Syntology official (archive's flag): 10 ran · 30 ran (of which 3 constructed an object rather than computing a result; 22 with no instrument failure: 2 honoured, 0 violated, 20 with no contract checked; 8 where Syntology's instrument failed) · 22 unverified Compare
CR (9 rows) AnglE-LLaMA-7B AnglE-optimized Text Embeddings code — Compare
SemEval 2014 Task 4 Subtask 1+2 (8 rows) InstructABSA InstructABSA: Instruction Learning for Aspect Based Sentiment Analysis code — Compare
SLUE (8 rows) W2V2-L-LL60K (pipeline approach, uses LM) SLUE: New Benchmark Tasks for Spoken Language Understanding... code Syntology official (archive's flag): 11 ran · 11 ran (of which 0 constructed an object rather than computing a result; 11 with no instrument failure: 0 honoured, 0 violated, 11 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified Compare
TweetEval (7 rows) BERTweet BERTweet: A pre-trained language model for English Tweets code — Compare
Multi-Domain Sentiment Dataset (6 rows) UDALM: Unsupervised Domain Adaptation through Language Modeling UDALM: Unsupervised Domain Adaptation through Language Modeling code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 5 with no instrument failure: 0 honoured, 0 violated, 5 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
DBRD (4 rows) RobBERT v2 RobBERT: a Dutch RoBERTa-based Language Model code Syntology official (archive's flag): 1 ran · 1 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
FiQA (4 rows) k-RoBERTa (parallel) Incorporating Multiple Knowledge Sources for Targeted Aspect-based... — — Compare
IITP Product Reviews Sentiment (4 rows) CalBERT CalBERT - Code-mixed Adaptive Language representations using BERT code — Compare
MPQA (4 rows) RoBERTa-large 355M + Entailment as Few-shot Learner Entailment as Few-Shot Learner code Syntology 1 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 1 violated, 0 with no contract checked; 0 where Syntology's instrument failed) · 2 unverified Compare
IITP Movie Reviews Sentiment (3 rows) xlmindic-base-uniscript Does Transliteration Help Multilingual Language Modeling? code Syntology official (archive's flag): 1 ran · 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified Compare
RuSentiment (3 rows) RuBERT-RuSentiment Deep Transfer Learning Baselines for Sentiment Analysis in Russian code — Compare
SemEval 2017 Task 4-A (3 rows) LSTMs+CNNs ensemble with multiple conv. ops BB_twtr at SemEval-2017 Task 4: Twitter Sentiment Analysis with... code — Compare
Twitter (3 rows) AEN-BERT Attentional Encoder Network for Targeted Sentiment Classification code — Compare
Financial PhraseBank (2 rows) FiLM Exploring the Impact of Corpus Diversity on Financial Pretrained... code — Compare
IMDb Movie Reviews (2 rows) Space-XLNet Breaking Free Transformer Models: Task-specific Context... code — Compare
SemEval (2 rows) MT-ISAO (Ours) Multi-Task Learning with LLMs for Implicit Sentiment Analysis:... — — Compare
122 People - Passenger Behavior Recognition Data (1 row) lstm+bert Deep Neural Networks for Bot Detection code Syntology 2 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 2 violated, 0 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
1B Words (1 row) Random Using millions of emoji occurrences to learn any-domain... code Syntology official (archive's flag): 1 ran · 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified Compare
AJGT (1 row) AraBERTv1 AraBERT: Transformer-based Model for Arabic Language Understanding code — Compare
ArSAS (1 row) CNN-LSTM Mazajak: An Online Arabic Sentiment Analyser — — Compare
ASTD (1 row) CNN-LSTM Mazajak: An Online Arabic Sentiment Analyser — — Compare
ChnSentiCorp (1 row) RoBERTa-wwm-ext-large Pre-Training with Whole Word Masking for Chinese BERT code — Compare
ChnSentiCorp Dev (1 row) RoBERTa-wwm-ext-large Pre-Training with Whole Word Masking for Chinese BERT code — Compare
HARD (1 row) AraBERTv1 AraBERT: Transformer-based Model for Arabic Language Understanding code — Compare
LABR (2-class, unbalanced) (1 row) AraBERTv1 AraBERT: Transformer-based Model for Arabic Language Understanding code — Compare
Latvian Twitter Eater Sentiment Dataset (1 row) Naive Bayes What Can We Learn From Almost a Decade of Food Tweets code — Compare
SAIL 2017 (1 row) CalBERT CalBERT - Code-mixed Adaptive Language representations using BERT code — Compare
Sogou News (1 row) fastText, h=10, bigram Bag of Tricks for Efficient Text Classification code Syntology official (archive's flag): 1 ran · 3 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 2 where Syntology's instrument failed) · 6 unverified Compare
Urdu Online Reviews (1 row) RCNN Sentiment analysis for Urdu online reviews using deep learning models code — Compare
Amazon Polarity (0 rows) no rows in the archive — —
BanglaSenti (0 rows) no rows in the archive — —
SetFit/tweet_sentiment_extraction (0 rows) no rows in the archive — —

Syntology column: samples harvested from the paper's repositories and executed on synthesized fixtures; “ran” is not a correctness claim and does not order the table. Each line reads “N ran (of which C constructed an object rather than computing a result; K with no instrument failure: H honoured, V violated, P with no contract checked; I where Syntology's instrument failed) · U unverified”; the instrument figure counts failures of Syntology's instrument, not of the code. When the archive marks a repository official for the row's paper, the line starts with that repository's state (“official” is the archive's flag, not a verdict on who wrote the code; “community repositories only” when every sample that ran came from a community repository, “official: no sample here; runs from other or unrecorded repositories” when some came from a repository the paper names or has in its text, or from none recorded); hover it for the repositories the samples that ran came from. A dash means no Syntology record for that paper, not a recorded non-run. Each “ran” line links to that paper's sample list. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. For agents: get_harvested_code_for_paper(arxiv_id) lists each paper's samples; how to connect.

Libraries

Not in the archive: the export carries no per-task library table, so there is nothing to show at snapshot 2025-07-28.

Datasets archive 2025-07-28

104 datasets whose archive record lists this task, ordered by the archive's paper count. 30 shown of 104 until expanded.

Subtasks archive 2025-07-28

12 subtasks in the archive's task tree.

Papers archive 2025-07-28

archive papers tagged: 5,630 · with a code link: 1,509 · where Syntology ran a sample: 220 (189 with a run with no instrument failure, 31 where every run was a failure of Syntology's instrument) Syntology

Show: all tagged papersonly where code ran (220 of 5,630 tagged: 189 with a run with no instrument failure, 31 where every run was a failure of Syntology's instrument)

Page 1 of 57: papers 1 to 100 of 5,630, in archive order: by repositories listed in the archive (most first), then newest first, not by stars (the archive holds no stars, so PwC's “Social” and “Latest” sorts cannot be reproduced). Papers that list no repository come after every paper that lists one.

Papers without a page here are shown as plain text. A Syntology line reads “N ran (of which C constructed an object rather than computing a result; K with no instrument failure: H honoured, V violated, P with no contract checked; I where Syntology's instrument failed) · U unverified”; the figure “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. When the archive marks a repository official for the paper, the line starts with that repository's state (the archive's flag, not a verdict on who wrote the code); hover it for the repositories the samples that ran came from. The first sentence of the abstract is shown for the first 30 papers; every abstract is on its paper's page.

Syntology lines on 62 of the papers shown; no Syntology record for the others (a paper without an arXiv id cannot be joined to the graph, and absence from the graph layer is not a recorded non-run). “Ran” means the sample executed on a synthesized fixture, not that the paper's result was reproduced; each line links to that paper's sample list. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. For agents: get_harvested_code_for_paper(arxiv_id) lists each paper's samples; how to connect.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections