R语言中提取自由文本数字及对应年龄单位的多组信息需求

核心思路 #

用stringr包的str_match_all提取所有匹配的年龄组合（数字+单位），通过正则捕获组拆分数字和单位，再整理成结构化数据；用tidyr的unnest可将多组年龄展开为多行，方便后续处理。

完整代码示例 #

# 加载依赖包
library(dplyr)
library(stringr)
library(tidyr)

# 示例数据集
dataset <- data.frame(
  Freetext = c(
    "A 3-year-old body was at the supermarket",
    "A 45 year old mother was at the supermarket with her little 5 months old boy",
    "The baby is 10 days-old and his grandma is 70 Jahre alt"
  ),
  Furtherinfo = c("ID1", "ID2", "ID3")
)

# 处理年龄提取与结构化
dataset_processed <- dataset %>%
  mutate(
    # 提取所有匹配的年龄组，捕获数字和单位
    age_matches = str_match_all(Freetext, 
                                "(\\d+)\\s*(?:(year|month|day)s?\\s*(?:old)?|-(year|month|day)-old|(year|month|day)s?-old|Jahre alt)") %>%
      lapply(function(x) {
        # 合并不同格式下的单位捕获组
        units <- coalesce(x[,2], x[,3], x[,4])
        # 统一德语单位为英文
        units <- ifelse(units == "Jahre", "year", units)
        # 返回结构化的年龄数据框
        data.frame(
          age_number = as.numeric(x[,1]),
          age_unit = units
        )
      })
  )

# 将多组年龄展开为多行（可选）
dataset_unnested <- dataset_processed %>%
  unnest(age_matches, keep_empty = TRUE)

# 查看结果
print(dataset_unnested)

代码说明 #

1. 正则表达式：覆盖了常见的年龄格式（如3-year-old、45 year old、5 months old、10 days-old、德语70 Jahre alt），通过捕获组分别提取数字和单位。
2. str_match_all：替代原代码的str_extract，支持提取文本中的多组年龄信息。
3. 单位统一：将德语的Jahre转换为year，方便后续统一分析。
4. unnest：把每个文本对应的多组年龄从列表列展开为多行，适合需要逐行分析单年龄的场景。

输出结果示例 #

Freetext Furtherinfo age_number age_unit
1                                A 3-year-old body was at the supermarket         ID1          3     year
2 A 45 year old mother was at the supermarket with her little 5 months old boy         ID2         45     year
3 A 45 year old mother was at the supermarket with her little 5 months old boy         ID2          5    month
4                 The baby is 10 days-old and his grandma is 70 Jahre alt         ID3         10      day
5                 The baby is 10 days-old and his grandma is 70 Jahre alt         ID3         70     year

内容的提问来源于stack exchange，提问作者USER12345