Lexicons and Tools for Italian Sentiment Analysis.

sentixr

R package to perform sentiment analysis on Italian texts, including four lexicons in tidy format (tibbles): Sentix, MAL, ELIta VAD and basic emotions (Plutchik’s wheel of emotions).

Overview

The package provides two main functions to perform sentiment analysis on Italian texts:

• sentix_annotate(): Annotates a text or a dataframe of texts with sentiment scores from the selected lexicon, using udpipe for parsing, and dplyr for joining the lexicon.
• sentix_summarize(): Summarizes the sentiment scores from the annotated dataframe, providing overall sentiment metrics per document (or segments) and auxiliary metrics that can be used to evaluate the results, recompute scores, or apply custom aggregation strategies.

It also provides utility functions to manage the included lexicons, allowing them to be easily used also within other frameworks such as Tidytext and Quanteda.

• get_sentix(), get_elita() and make_polarity(), to extract and use them in custom analyses;
• df_to_valence(), df_to_polar() (plus the wrapper df_to_dict()) to convert tidy lexicons into Quanteda dictionaries.

Install

remotes::install_github("valeriobasile/sentixr", dependencies = TRUE, 
                         build_vignettes = TRUE)

Lexicons

While the package is published under a GPL-3.0 license, lexicons are provided under separate terms (CC BY-SA and CC0 1.0 Universal).

Please check the individual lexicon documentation for details.

Usage

library(sentixr)

Main workflow

The sentiment analysis workflow consists of two main steps: annotation (sentix_annotate) and summarization (sentix_summarize).

# Single text
# annotate with defaults
sentix_annotate("Oggi è una bella giornata. Esco a fare una passeggiata") |>
  # summarize by document - default
  sentix_summarize()

# A tibble: 1 × 4
  doc_id score n_tokens n_scored
  <chr>  <dbl>    <int>    <int>
1 doc1   0.176       10        9

Annotate

sentix_annotate() annotates texts with sentiment scores from the selected lexicon, after parsing them with udpipe. For large corpora, the user may optionally specify the number of cores to use, via the argument parallel.cores, which is inherited from udpipe and passed to udpipe::udpipe().

Managing the udpipe model

The model argument allows specifying a custom udpipe model. If no model is given, the function will automatically download the default Italian udpipe model. After the first run, the downloaded model can be reused.

# Using a model in the working directory
sentix_annotate(testi, model = "local")

# Loading a pre-downloaded model
model <- udpipe::udpipe_load_model("italian-isdt-ud-2.5-191206.udpipe")

With multiple texts and dataframe

The function, like udpipe, accepts as input single texts, multiple texts (a character vector, a list or a list of tokens), or dataframes with text and doc_id columns.

The function also helps the user specify custom column names and manage document identifiers, which are safely passed to udpipe::udpipe(). Note that, however, other columns in the dataframe are ignored.

# Example dataframe with doc_id and text fields
data(recensioni_tv)

# Annotate the dataframe directly}
anno_df <- sentix_annotate(
  recensioni_tv,
  # loaded model
  model = model
)

head(anno_df)

  doc_id sentence_id token_id    token    lemma  upos     score
1   doc1           1        1   Ottimo   ottimo   ADJ 1.0000000
2   doc1           1        2 prodotto prodotto  NOUN 0.0000000
3   doc1           1        3        ,        , PUNCT        NA
4   doc1           1        4       la       il   DET        NA
5   doc1           1        5  qualità  qualità  NOUN 0.3631757
6   doc1           1      6-7    dell'     <NA>  <NA>        NA

With a different lexicon

By default, sentix_annotate() uses the Sentix lexicon. To use a different lexicon, specify it with the dict argument.

# Use ELIta lexicon with VAD scores
anno_vad <- sentix_annotate(recensioni_tv, model = model, dict = "elita_VAD")

Summarize

sentix_summarize() computes overall sentiment scores and auxiliary metrics per document (or other segments, via the argument by) from the annotated dataframe.

sentix_summarize(
  anno_df,
  # summarize by sentence
  by = c("doc_id", "sentence_id")
)

# A tibble: 7 × 5
  doc_id sentence_id   score n_tokens n_scored
  <chr>        <int>   <dbl>    <int>    <int>
1 doc1             1  0.274        12        9
2 doc2             1 -0.253         4        3
3 doc2             2 -0.0818       11        6
4 doc3             1  0.244        15        9
5 doc4             1  0.178         4        3
6 doc4             2  0.0965       12        9
7 doc5             1 -0.0187       15        9

When using lexicons with multiple features (e.g., ELIta), all features are summarized.

sentix_summarize(anno_vad)

# A tibble: 5 × 6
  doc_id  valenza attivazione dominanza n_tokens n_scored
  <chr>     <dbl>       <dbl>     <dbl>    <int>    <int>
1 doc1    0.500       -0.0971    0.285        12        6
2 doc2    0.0414       0.387     0.0957       15        7
3 doc3    0.402        0.208     0.125        15        3
4 doc4    0.0714      -0.0296    0.0954       16        7
5 doc5   -0.00667      0.0554    0.0342       15        6

How to Cite

If you use sentixr in your research, please cite it as follows:

Vardanega, A., Basile, V., Vassallo, M., Gabrieli, G. & Di Palma, E. (2026). sentixr (Versione 0.1.0) https://github.com/valeriobasile/sentixR

Authors

• Vardanega, Agnese (Università di Teramo) (mantainer)

• Basile, Valerio (Università di Torino)

• Vassallo, Marco (CREA-PB)

• Gabrieli, Giuliano (CREA-PB)

• Di Palma, Eliana (Università di Torino)

References