ragnar_experiment.R

library(tidyverse)
library(ragnar)

# creating a store --------------------------------------------------------
ragnar_store_create(
  location = "english_linguistic_abstract",
  overwrite = TRUE,
  version = 1,
  embed = function(x){ragnar::embed_ollama(x, model = "embeddinggemma")},
  extra_cols = data.frame(author = character(),
                          title = character(),
                          journal = character(),
                          volume = character(),
                          issue = charactezr(),
                          publication_year = character(),
                          first_page = character(),
                          last_page = character(),
                          doi = character())) ->
  store

read_csv("data/2024-04-22.csv", show_col_types = FALSE) |>  # путь к одному файлу с аннотациями
  select(doi, author, title, publication_year, first_page, last_page, volume, issue, journal_issn, abstract) |> 
  rename(journal = journal_issn,
         text = abstract) |> 
  mutate_all(as.character) |> 
  mutate_all(function(x) ifelse(is.na(x), "-", x)) |> 
  ragnar_store_insert(store, chunks = _)

ragnar_store_build_index(store)

# retrieving from the store -----------------------------------------------

store <- ragnar_store_connect("english_linguistic_abstract")

ragnar_store_inspect(store)

ragnar_retrieve_vss(store,
                    method = "cosine_distance", 
                    deoverlap = FALSE, 
                    top_k = 20,
                    query = "number",
                    filter = author == "Akira Watanabe") |> 
  select(-embedding, -hash, -metric_name, -origin) |> 
  relocate(text, .before = "author") |> 
  relocate(title, .before = "author") |> 
  arrange(metric_value) |> 
  View()