R语言中矩阵转Dataframe失败问题求助

我编写了一个调用OpenTapioca API并解析标注的id、label、description和score的函数，但将数据转换为DataFrame时出现异常。目前data_matrix矩阵显示正常，但转换为DataFrame后结果不对。

原函数代码 #

get_wikidata_links <- function(input_text, minimum_score) {
  
  #
  # Function which takes a character vector of length 1 as input (i.e. all text
  # needs to be combined into a single character) as well as a minimum certainty
  # score, and returns a tibble with key information and links to Wikidata
  #
  # Input
  #  - input_text: Text input (character)
  #  - minimum_score: Minimum score that every returned entity needs to have
  #                   (numeric)
  #
  # Output
  #  - top_wikidata_links: Table with the first four columns being 'id', 'label',
  #               'description', 'score' (tibble)
  #
  
  base_url <- "https://opentapioca.org/api/annotate"
  r <- GET(base_url, query = list(query = input_text))
  
  
  data = content(r)$annotations
  
  framed = list()
  vec = list()
  dummy = 0
  for (i in 1:length(data)) {
    
    data1 = data[[i]]$tags
    
    for (j in 1:length(data1)) {
      
      data2 = data1[[j]]

      if (data2$score>minimum_score) {
        
        vec[1] <- data2$id
        vec[2] <- data2$label
        vec[3] <- data2$desc
        vec[4] <- data2$score
        dummy <-  dummy + 1
        framed[[dummy]] <- vec
      }
    }
  }
  
  data_matrix <- do.call("rbind", framed)
  top_wikidata_links <- as.data.frame(data_matrix, stringsAsFactors = FALSE)
  colnames(top_wikidata_links) <- c("ID", "Label", "Description", "Score")
  
  return(top_wikidata_links)
  
}

测试代码 #

# Test 1
text_example_1 <- c("Karl Popper worked at the LSE.")
get_wikidata_links(input_text_1, -0.5) # 注：原代码参数名错误，应为text_example_1
# 
# Hint: The output should be a tibble similar to the one outlined below
#
# | id | label | description | score |
# | "Q81244" | "Karl Popper" | "Austrian-British philosopher of science" | 2.4568285 |
# | "Q174570" | "London School of Economics and Political Science" | "university in Westminster, UK" | "1.4685043" |
# | "Q171240" | "London Stock Exchange" | "stock exchange in the City of London" | "-0.4124461" |

# Test 2
text_example_2 <- c("Claude Shannon studied at the University of Michigan and at MIT.")
get_wikidata_links(text_example_2, 0)

问题根源 #

原代码中vec列表是在循环外部定义的，每次循环只是修改这个列表的元素，而非创建新的列表对象。这导致framed中的所有元素都指向同一个vec，最终所有行都会是最后一次循环的结果，引发DataFrame异常。此外，手动转矩阵再转DataFrame的方式容易丢失数据类型（比如Score会被转为字符型），且效率较低。

修复方案 #

方案1：修复原循环逻辑 #

每次符合条件时创建新的列表，避免引用同一对象，同时用bind_rows直接生成tibble：

library(httr)
library(tibble)

get_wikidata_links <- function(input_text, minimum_score) {
  
  base_url <- "https://opentapioca.org/api/annotate"
  r <- GET(base_url, query = list(query = input_text))
  
  data <- content(r)$annotations
  
  framed <- list()
  dummy <- 0
  for (i in 1:length(data)) {
    
    data1 <- data[[i]]$tags
    
    for (j in 1:length(data1)) {
      
      data2 <- data1[[j]]
      
      if (data2$score > minimum_score) {
        # 每次创建独立的列表，避免共享引用
        vec <- list(
          ID = data2$id,
          Label = data2$label,
          Description = data2$desc,
          Score = data2$score
        )
        dummy <- dummy + 1
        framed[[dummy]] <- vec
      }
    }
  }
  
  # 处理无符合条件结果的情况
  if (length(framed) == 0) {
    return(tibble(ID = character(), Label = character(), Description = character(), Score = numeric()))
  }
  
  top_wikidata_links <- bind_rows(framed)
  return(top_wikidata_links)
}

方案2：用tidyverse简化代码（推荐） #

利用purrr包的扁平化和映射操作，避免嵌套循环，代码更简洁可靠：

library(httr)
library(tibble)
library(purrr)

get_wikidata_links <- function(input_text, minimum_score) {
  
  base_url <- "https://opentapioca.org/api/annotate"
  r <- GET(base_url, query = list(query = input_text))
  
  content(r)$annotations %>%
    map("tags") %>% # 提取每个annotation下的tags列表
    flatten() %>% # 扁平化嵌套列表
    keep(~ .x$score > minimum_score) %>% # 过滤分数达标项
    map_dfr(~ tibble( # 映射并合并为tibble
      ID = .x$id,
      Label = .x$label,
      Description = .x$desc,
      Score = .x$score
    ))
}

修正后的测试代码 #

# Test 1
text_example_1 <- c("Karl Popper worked at the LSE.")
get_wikidata_links(text_example_1, -0.5)

# Test 2
text_example_2 <- c("Claude Shannon studied at the University of Michigan and at MIT.")
get_wikidata_links(text_example_2, 0)

内容的提问来源于stack exchange，提问作者Jay