opengraph

Tools to Mine ‘Open Graph’-like Tags From ‘HTML’ Content

Description

Many services such as ‘Facebook’ and ‘Twitter’ rely on ‘Open Graph’-like tags (http://ogp.me/) to build a formalized knowledge graph from ‘HTML’ content metadata. Tools are providfed to extract and analyze this data.

What’s Inside The Tin

The following functions are implemented:

• read_properties: Read Open Graph tags from an HTML document
• is_valid_property: Is an property valid?

Installation

devtools::install_github("hrbrmstr/opengraph")

Usage

library(opengraph)
library(jsonlite)
library(hrbrthemes)
library(tidyverse)

# current verison
packageVersion("opengraph")

## [1] '0.1.0'

Tag Distribution

rt <- rprojroot::find_package_root_file()

if (!file.exists(file.path(rt, "ex", "tags.rds"))) {

# Hacker News current top 500 ~1700 EST 2017-12-24
x <- jsonlite::fromJSON("https://hacker-news.firebaseio.com/v0/topstories.json")

# Get the metadat for each one
pb <- progress_estimated(length(x))
sprintf("https://hacker-news.firebaseio.com/v0/item/%s.json", x) %>% 
  map(~{
    pb$tick()$print()
    jsonlite::fromJSON(.x)
  }) -> stories

# Deal with any HTTP or parsing errors
s_read_props <- purrr::safely(opengraph::read_properties)

# Go to each site & pull the metadata from it
map(stories, "url") %>% 
  flatten_chr() %>% 
  map_df(~{
    cat(".")
    res <- s_read_props(.x) 
    if (is.null(res$result)) return(data_frame(orig_url = .x))
    mutate(res$result, orig_url = .x)
  }) -> tags

# Don't waste bandwidth
write_rds(tags, file.path(rt, "ex", "tags.rds"))

} else {
  tags <- read_rds(file.path(rt, "ex", "tags.rds"))
}

# Which ones are valid?
tags <- mutate(tags, is_valid = is_valid_property(property))

filter(tags, is_valid) %>% 
  count(is_valid, orig_url) -> valid_sites

nsites <- length(unique(valid_sites$orig_url))

ggplot() +
  ggalt::geom_bkde(data=valid_sites, aes(n, y=..count..), color="#3f007d", fill="#3f007d", size=0.5, alpha=3/5, bandwidth=1) +
  geom_rug(data=valid_sites, aes(n), color="#3f007d", size=0.25) +
  scale_x_comma(limits=c(0, NA), 
                breaks=sort(c(1, seq(0, 90, 10))),
                labels=c("\n0", 1, seq(0, 90, 10)[-1])) +
  scale_y_continuous(expand=c(0,1)) +
  labs(x="# OpenGraph tags on a site",
       y="# Sites", 
       title="Distribution of the number of Open Graph-ish tags per-site\nfrom 1st 500 Hacker News Article URLs",
       subtitle=sprintf("%s of 500 sites (%s) had Open Graph-ish tags", nsites, scales::percent(nsites/500)),
       caption="(Firebase pull & crawl start timestamp: 2017-12-24 ~1700EST)") +
  theme_ipsum_ps(grid="XY")

Top 20 Tags

filter(tags, is_valid) %>% 
  count(property, sort=TRUE) %>% 
  head(20) %>% 
  knitr::kable()

property	n
og:title	348
og:image	341
og:description	335
og:url	331
og:type	320
fb:pages	302
og:site_name	288
article:tag	249
fb:app_id	166
article:published_time	114
article:author	84
og:locale	83
og:image:width	78
og:image:height	77
og:see_also	70
article:modified_time	66
article:section	65
fb:admins	62
og:updated_time	30
twitter:image	27