Getting Data from the Web Scraping

You’ve tried everything else, and you haven’t managed to get your hands on the data you want. You’ve found the data on the web, but, alas — no download options are available and copy-paste has failed you. Fear not, there may still be a way to get the data out. For example you can:

•    Get data from web-based APIs, such as interfaces provided by online databases and many modern web applications (including Twitter, Facebook and many others). This is a fantastic way to access government or commercial data, as well as data from social media sites.

•    Extract data from PDFs. This is very difficult, as PDF is a language for printers and does not retain much information on the structure of the data that is displayed within a document. Extracting information from PDFs is beyond the scope of this book, but there are some tools and tutorials that may help you do it.

•    Screen scrape web sites. During screen scraping, you’re extracting structured content from a normal web page with the help of a scraping utility or by writing a small piece of code. While this method is very powerful and can be used in many places, it requires a bit of understanding about how the web works.

With all those great technical options, don’t forget the simple options: often it is worth to spend some time searching for a file with machine-readable data or to call the institution which is holding the data you want.

In this chapter we walk through a very basic example of scraping data from an HTML web page.

What is machine-readable data?

The goal for most of these methods is to get access to machine-readable data. Machine readable data is created for processing by a computer, instead of the presentation to a human user. The structure of such data relates to contained information, and not the way it is displayed eventually. Examples of easily machine-readable formats include CSV, XML, JSON and Excel files, while formats like Word documents, HTML pages and PDF files are more concerned with the visual layout of the information. PDF for example is a language which talks directly to your printer, it’s concerned with position of lines and dots on a page, rather than distinguishable characters.

Scraping web sites: what for?

Everyone has done this: you go to a web site, see an interesting table and try to copy it over to Excel so you can add some numbers up or store it for later. Yet this often does not really work, or the information you want is spread across a large number of web sites. Copying by hand can quickly become very tedious, so it makes sense to use a bit of code to do it.

The advantage of scraping is that you can do it with virtually any web site — from weather forecasts to government spending, even if that site does not have an API for raw data access.

What you can and cannot scrape

There are, of course, limits to what can be scraped. Some factors that make it harder to scrape a site include:

•    Badly formatted HTML code with little or no structural information e.g. older government websites.

•    Authentication systems that are supposed to prevent automatic access e.g. CAPTCHA codes and paywalls.

•    Session-based systems that use browser cookies to keep track of what the user has been doing.

•    A lack of complete item listings and possibilities for wildcard search.

•    Blocking of bulk access by the server administrators.

Another set of limitations are legal barriers: some countries recognize database rights, which may limit your right to re-use information that has been published online. Sometimes, you can choose to ignore the license and do it anyway — depending on your jurisdiction, you may have special rights as a journalist. Scraping freely available Government data should be fine, but you may wish to double check before you publish. Commercial organizations — and certain NGOs — react with less tolerance and may try to claim that you’re “sabotaging” their systems. Other information may infringe the privacy of individuals and thereby violate data privacy laws or professional ethics.

Tools that help you scrape

There are many programs that can be used to extract bulk information from a web site, including browser extensions and some web services. Depending on your browser, tools like Readability (which helps extract text from a page) or DownThemAll (which allows you to download many files at once) will help you automate some tedious tasks, while Chrome’s Scraper extension was explicitly built to extract tables from web sites. Developer extensions like FireBug (for Firefox, the same thing is already included in Chrome, Safari and IE) let you track exactly how a web site is structured and what communications happen between your browser and the server.

ScraperWiki is a web site that allows you to code scrapers in a number of different programming languages, including Python, Ruby and PHP. If you want to get started with scraping without the hassle of setting up a programming environment on your computer, this is the way to go. Other web services, such as Google Spreadsheets and Yahoo! Pipes also allow you to perform some extraction from other web sites.

How does a web scraper work?

Web scrapers are usually small pieces of code written in a programming language such as Python, Ruby or PHP. Choosing the right language is largely a question of which community you have access to: if there is someone in your newsroom or city already working with one of these languages, then it makes sense to adopt the same language.

While some of the click-and-point scraping tools mentioned before may be helpful to get started, the real complexity involved in scraping a web site is in addressing the right pages and the right elements within these pages to extract the desired information. These tasks aren’t about programming, but understanding the structure of the web site and database.

When displaying a web site, your browser will almost always make use of two technologies: HTTP is a way for it to communicate with the server and to request specific resource, such as documents, images or videos. HTML is the language in which web sites are composed.

The anatomy of a web page

Any HTML page is structured as a hierarchy of boxes (which are defined by HTML “tags”). A large box will contain many smaller ones — for example a table that has many smaller divisions: rows and cells. There are many types of tags that perform different functions — some produce boxes, others tables, images or links. Tags can also have additional properties (e.g. they can be unique identifiers) and can belong to groups called ‘classes’, which makes it possible to target and capture individual elements within a document. Selecting the appropriate elements this way and extracting their content is the key to writing a scraper.

Viewing the elements in a web page: everything can be broken up into boxes within boxes.

To scrape web pages, you’ll need to learn a bit about the different types of elements that can be in an HTML document. For example, the <table> element wraps a whole table, which has <tr> (table row) elements for its rows, which in turn contain <td> (table data) for each cell. The most common element type you will encounter is <div>, which can basically mean any block of content. The easiest way to get a feel for these elements is by using the developer toolbar in your browser: they will allow you to hover over any part of a web page and see what the underlying code is.

Tags work like book ends, marking the start and the end of a unit. For example <em> signifies the start of an italicized or emphasized piece of text and </em> signifies the end of that section. Easy.

Figure 57. The International Atomic Energy Agency’s (IAEA) portal (news.iaea.org)

An example: scraping nuclear incidents with Python

NEWS is the International Atomic Energy Agency’s (IAEA) portal on world-wide radiation incidents (and a strong contender for membership in the Weird Title Club!). The web page lists incidents in a simple, blog-like site that can be easily scraped.

To start, create a new Python scraper on ScraperWiki and you will be presented with a text area that is mostly empty, except for some scaffolding code. In another browser window, open the IAEA site and open the developer toolbar in your browser. In the “Elements” view, try to find the HTML element for one of the news item titles. Your browser’s developer toolbar helps you connect elements on the web page with the underlying HTML code.

Investigating this page will reveal that the titles are <h4> elements within a <table>. Each event is a <tr> row, which also contains a description and a date. If we want to extract the titles of all events, we should find a way to select each row in the table sequentially, while fetching all the text within the title elements.

In order to turn this process into code, we need to make ourselves aware of all the steps involved. To get a feeling for the kind of steps required, let’s play a simple game: In your ScraperWiki window, try to write up individual instructions for yourself, for each thing you are going to do while writing this scraper, like steps in a recipe (prefix each line with a hash sign to tell Python that this not real computer code). For example:

# Look for all rows in the table

# Unicorn must not overflow on left side.

Try to be as precise as you can and don’t assume that the program knows anything about the page you’re attempting to scrape.

Once you’ve written down some pseudo-code, let’s compare this to the essential code for our first scraper:

import scraperwiki

In this first section, we’re importing existing functionality from libraries — snippets of pre-written code. scraperwiki will give us the ability to download web sites, while lxml is a tool for the structured analysis of HTML documents. Good news: if you are writing a Python scraper with ScraperWiki, these two lines will always be the same.

doc_text = scraperwiki.scrape(url)

doc = html.fromstring(doc_text)

Next, the code makes a name (variable): url, and assigns the URL of the IAEA page as its value. This tells the scraper that this thing exists and we want to pay attention to it. Note that the URL itself is in quotes as it is not part of the program code but a string, a sequence of characters.

We then use the url variable as input to a function, scraperwiki.scrape. A function will provide some defined job — in this case it’ll download a web page. When it’s finished, it’ll assign its output to another variable, doc_text. doc_text will now hold the actual text of the website — not the visual form you see in your browser, but the source code, including all the tags. Since this form is not very easy to parse, we’ll use another function, html.fromstring, to generate a special representation where we can easily address elements, the so-called document object model (DOM).

In this final step, we use the DOM to find each row in our table and extract the event’s title from its header. Two new concepts are used: the for loop and element selection (.cssselect). The for loop essentially does what its name implies; it will traverse a list of items, assigning each a temporary alias (row in this case) and then run any indented instructions for each item.

The other new concept, element selection, is making use of a special language to find elements in the document. CSS selectors are normally used to add layout information to HTML elements and can be used to precisely pick an element out of a page. In this case (Line. 6) we’re selecting #tblEvents tr which will match each <tr> within the table element with the ID tblEvents (the hash simply signifies ID). Note that this will return a list of <tr> elements.

As can be seen on the next line (Line. 7), where we’re applying another selector to find any <a> (which is a hyperlink) within a <h4> (a title). Here we only want to look at a single element (there’s just one title per row), so we have to pop it off the top of the list returned by our selector with the .pop() function.

Note that some elements in the DOM contain actual text, i.e. text that is not part of any markup language, which we can access using the [element].text syntax seen on line 8. Finally, in line 9, we’re printing that text to the ScraperWiki console. If you hit run in your scraper, the smaller window should now start listing the event’s names from the IAEA web site.

You can now see a basic scraper operating: it downloads the web page, transforms it into the DOM form and then allows you to pick and extract certain content. Given this skeleton, you can try and solve some of the remaining problems using the ScraperWiki and Python documentation:

•    Can you find the address for the link in each event’s title?

•    Can you select the small box that contains the date and place by using its CSS class name and extract the element’s text?

•    ScraperWiki offers a small database to each scraper so you can store the results; copy the relevant example from their docs and adapt it so it will save the event titles, links and dates.

•    The event list has many pages; can you scrape multiple pages to get historic events as well?

As you’re trying to solve these challenges, have a look around ScraperWiki: there are many useful examples in the existing scrapers — and quite often, the data is pretty exciting, too. This way, you don’t need to start off your scraper from scratch: just choose one that is similar, fork it and adapt to your problem.

Source: http://datajournalismhandbook.org/1.0/en/getting_data_3.html

Data Scraping Services - Scraping Yelp Business Data With Python Scraping Script

Yelp is a great source of business contact information with details like address, postal code, contact information; website addresses etc. that other site like Google Maps just does not. Yelp also provides reviews about the particular business. The yelp business database can be useful for telemarketing, email marketing and lead generation.

Are you looking for yelp business details database? Are you looking for scraping data from yelp website/business directory? Are you looking for yelp screen scraping software? Are you looking for scraping the business contact information from the online Yelp? Then you are at the right place.

Here I am going to discuss how to scrape yelp data for lead generation and email marketing. I have made a simple and straight forward yelp data scraping script in python that can scrape data from yelp website. You can use this yelp scraper script absolutely free.

I have used urllib, BeautifulSoup packages. Urllib package to make http request and parsed the HTML using BeautifulSoup, used Threads to make the scraping faster.

Yelp Scraping Python Script

import urllib from bs4 import BeautifulSoup import re from threading import Thread #List of yelp urls to scrape url=['http://www.yelp.com/biz/liman-fisch-restaurant-hamburg','http://www.yelp.com/biz/casa-franco-caramba-hamburg','http://www.yelp.com/biz/o-ren-ishii-hamburg','http://www.yelp.com/biz/gastwerk-hotel-hamburg-hamburg-2','http://www.yelp.com/biz/superbude-hamburg-2','http://www.yelp.com/biz/hotel-hafen-hamburg-hamburg','http://www.yelp.com/biz/hamburg-marriott-hotel-hamburg','http://www.yelp.com/biz/yoho-hamburg'] i=0 #function that will do actual scraping job def scrape(ur): html = urllib.urlopen(ur).read() soup = BeautifulSoup(html) title = soup.find('h1',itemprop="name") saddress = soup.find('span',itemprop="streetAddress") postalcode = soup.find('span',itemprop="postalCode") print title.text print saddress.text print postalcode.text print "-------------------" threadlist = [] #making threads while i<len(url): t = Thread(target=scrape,args=(url[i],)) t.start() threadlist.append(t) i=i+1 for b in

threadlist: b.join()

import urllib

from bs4 import BeautifulSoup

import re

from threading import Thread

 #List of yelp urls to scrape

url=['http://www.yelp.com/biz/liman-fisch-restaurant-hamburg','http://www.yelp.com/biz/casa-franco-caramba-hamburg','http://www.yelp.com/biz/o-ren-ishii-hamburg','http://www.yelp.com/biz/gastwerk-hotel-hamburg-hamburg-2','http://www.yelp.com/biz/superbude-hamburg-2','http://www.yelp.com/biz/hotel-hafen-hamburg-hamburg','http://www.yelp.com/biz/hamburg-marriott-hotel-hamburg','http://www.yelp.com/biz/yoho-hamburg']

 i=0

#function that will do actual scraping job

def scrape(ur):

           html = urllib.urlopen(ur).read()

          soup = BeautifulSoup(html)

       title = soup.find('h1',itemprop="name")

          saddress = soup.find('span',itemprop="streetAddress")

          postalcode = soup.find('span',itemprop="postalCode")

          print title.text

          print saddress.text

          print postalcode.text

          print "-------------------"

 threadlist = []

#making threads

while i<len(url):

          t = Thread(target=scrape,args=(url[i],))

          t.start()

          threadlist.append(t)

          i=i+1

for b in threadlist:

          b.join()

Recently I had worked for one German company and did yelp scraping project for them and delivered data as per their requirement. If you looking for scraping data from business directories like yelp then send me your requirement and I will get back to you with sample.

Source: http://webdata-scraping.com/scraping-yelp-business-data-python-scraping-script/

Endorsing web scraping

With more than 200 projects delivered, we stand firmly for new challenges every day. We have served above 60 clients and have won 86% of repeat business, as our main core is customer delight. Successive Softwares was approached by a client having a very exclusive set of requirements. For their project they required customised data mining, in real time to offer profitable information to their customers. Requirement stated scrapping of stock exchange data in real time so that end users can be eased in their marketing decisions. This posed as an ambitious task for us because it required processing of huge amount of data on a routine basis. We welcomed it as an event to evolve and do something aside of classic web application development.

We started with mock-ups, pursuing our very first step of IMPART Framework (Innovative Mock-up based Prototypes Analyzed to develop Reengineered Technology). Our team of experts thought of all the potential requirements with a flow and materialized it flawlessly into our mock up. It was a strenuous tasks but our excitement to do something which others still do not think of, filled our team with confidence and energy and things began to roll out perfectly. We presented our mock-up and statistics to the client as per our expectation client choose us, impressed with the efforts.

We started gathering requirements from client side and started to formulate design about the flow. The project required real time monitoring of stock exchange together with Prices, Market Turnover and then implement them into graphs. The front end part was an easy deal, we were already adept in playing with data the way required. The intractable task was to get the data. We researched and found that it can be achieved either with API or with Web Scarping and we moved with latter because of the limitations in API.

Web scraping is a compelling technique to get the required information straight out of the web page. Lack of documentation and not much forbearance forced us to make a slow start, but we kept all the requirements clear and new that we headed in the right direction.  We divided the scraping process into bits of different but related tasks. Firstly we needed to find the data which has to be captured, some of the problems faced were pagination and use of AJAX but with examination of endpoints in URL and the requests made when data is drawn, we surmounted these problems easily.

After targeting our data we focused on HTML parser which could extract data form all the targets. Using PHP we developed a parser extracting all the information and saving them in Database in a structured way.  After the required data present at our end we easily manipulated it into tables and charts and we used HIGHSTOCK for that. Entire Client side was developed in PHP with Zend frame work and we used MySQL 5.7 for server side.

During the whole development cycle our QA team insured we were delivering a quality product following all standards. We kept our client in the loop during the whole process keeping them informed about every step. Clients were also assured as they watched their project starting from scratch which developed into full fledge website. The process followed a strict time line releasing regular builds and implementing new improvements. We stood up to the expectation our client and delivered a product just as they visualized it to be.

Source: http://www.successivesoftwares.com/endorsing-web-scraping/

What you need to know about web scraping: How to understand, identify, and sometimes stop

NB: This is a gust article by Rami Essaid, co-founder and CEO of Distil Networks.

Here’s the thing about web scraping in the travel industry: everyone knows it exists but few know the details.

Details like how does web scraping happen and how will I know? Is web scraping just part of doing business online, or can it be stopped? And lastly, if web scraping can be stopped, should it always be stopped?

These questions and the challenge of web scraping are relevant to every player in the travel industry. Travel suppliers, OTAs and meta search sites are all being scraped. We have the data to prove it; over 30% of travel industry website visitors are web scrapers.

Google Analytics, and most other analytics tools do not automatically remove web scraper traffic, also called “bot” traffic, from your reports – so how would you know this non-human and potentially harmful traffic exists? You have to look for it.

This is a good time to note that I am CEO of a bot-blocking company called Distil Networks, and we serve the travel industry as well as digital publishers and eCommerce sites to protect against web scraping and data theft – we’re on a mission to make the web more secure.

So I am admittedly biased, but will do my best to provide an educational account of what we’ve learned to be true about web scraping in travel – and why this is an issue every travel company should at the very least be knowledgeable about.

Overall, I see an alarming lack of awareness around the prevalence of web scraping and bots in travel, and I see confusion around what to do about it. As we talk this through I’ll explain what these “bots” are, how to find them and how to manage them to better protect and leverage your travel business.

What are bots, web scrapers and site indexers? Which are good and which are bad?

The jargon around web scraping is confusing – bots, web scrapers, data extractors, price scrapers, site indexers and more – what’s the difference? Allow me to quickly clarify.

–> Bots: This is a general term that refers to non-human traffic, or robot traffic that is computer generated. Bots are essentially a line of code or a program that is created to perform specific tasks on a large scale.  Bots can include web scrapers, site indexers and fraud bots. Bots can be good or bad.

–> Web Scraper: (web harvesting or web data extraction) is a computer software technique of extracting information from websites (source, Wikipedia). Web scrapers are usually bad.

If your travel website is being scraped, it is most likely your competitors are collecting competitive intelligence on your prices. Some companies are even built to scrape and report on competitive price as a service. This is difficult to prove, but based on a recent Distil Networks study, prices seem to be main target.You can see more details of the study and infographic here.

One case study is Ryanair. They have been particularly unhappy about web scraping and won a lawsuit against a German company in 2008, incorporated Captcha in 2011 to stop new scrapers, and when Captcha wasn’t totally effective and Cheaptickets was still scraping, they took to the courts once again.

So Ryanair is doing what seems to be a consistent job of fending off web scrapers – at least after the scraping is performed. Unfortunately, the amount of time and energy that goes into identifying and stopping web scraping after the fact is very high, and usually this means the damage has been done.

This type of web scraping is bad because:

    Your competition is likely collecting your price data for competitive intelligence.

    Other travel companies are collecting your flights for resale without your consent.

    Identifying this type of web scraping requires a lot of time and energy, and stopping them generally requires a lot more.

Web scrapers are sometimes good

Sometimes a web scraper is a potential partner in disguise.

Meta search sites like Hipmunk sometimes get their start by scraping travel site data. Once they have enough data and enough traffic to be valuable they go to suppliers and OTAs with a partnership agreement. I’m naming Hipmunk because the Company is one of th+e few to fess up to site scraping, and one of the few who claim to have quickly stopped scraping when asked.

I’d wager that Hipmunk and others use(d) web scraping because it’s easy, and getting a decision maker at a major travel supplier on the phone is not easy, and finding legitimate channels to acquire supplier data is most definitely not easy.

I’m not saying you should allow this type of site scraping – you shouldn’t. But you should acknowledge the opportunity and create a proper channel for data sharing. And when you send your cease and desist notices to tell scrapers to stop their dirty work, also consider including a note for potential partners and indicate proper channels to request data access.

–> Site Indexer: Good.

Google, Bing and other search sites send site indexer bots all over the web to scour and prioritize content. You want to ensure your strategy includes site indexer access. Bing has long indexed travel suppliers and provided inventory links directly in search results, and recently Google has followed suit.

–> Fraud Bot: Always bad.

Fraud bots look for vulnerabilities and take advantage of your systems; these are the pesky and expensive hackers that game websites by falsely filling in forms, clicking ads, and looking for other vulnerabilities on your site. Reviews sections are a common attack vector for these types of bots.

How to identify and block bad bots and web scrapers

Now that you know the difference between good and bad web scrapers and bots, how do you identify them and how do you stop the bad ones? The first thing to do is incorporate bot-identification into your website security program. There are a number of ways to do this.

In-house

When building an in house solution, it is important to understand that fighting off bots is an arms race. Every day web scraping technology evolves and new bots are written. To have an effective solution, you need a dynamic strategy that is always adapting.

When considering in-house solutions, here are a few common tactics:

    CAPTCHAs – Completely Automated Public Turing Tests to Tell Computers and Humans Apart (CAPTCHA), exist to ensure that user input has not been generated by a computer. This has been the most common method deployed because it is simple to integrate and can be effective, at least at first. The problem is that Captcha’s can be beaten with a little workand more importantly, they are a nuisance to end usersthat can lead to a loss of business.

    Rate Limiting- Advanced scraping utilities are very adept at mimicking normal browsing behavior but most hastily written scripts are not. Bots will follow links and make web requests at a much more frequent, and consistent, rate than normal human users. Limiting IP’s that make several requests per second would be able to catch basic bot behavior.

    IP Blacklists - Subscribing to lists of known botnets & anonymous proxies and uploading them to your firewall access control list will give you a baseline of protection. A good number of scrapers employ botnets and Tor nodes to hide their true location and identity. Always maintain an active blacklist that contains the IP addresses of known scrapers and botnets as well as Tor nodes.

    Add-on Modules – Many companies already own hardware that offers some layer of security. Now, many of those hardware providers are also offering additional modules to try and combat bot attacks. As many companies move more of their services off premise, leveraging cloud hosting and CDN providers, the market share for this type of solution is shrinking.

    It is also important to note that these types of solutions are a good baseline but should not be expected to stop all bots. After all, this is not the core competency of the hardware you are buying, but a mere plugin.

Some example providers are:

    Impreva SecureSphere- Imperva offers Web Application Firewalls, or WAF’s. This is an appliance that applies a set of rules to an HTTP connection. Generally, these rules cover common attacks such as Cross-site Scripting (XSS) and SQL Injection. By customizing the rules to your application, many attacks can be identified and blocked. The effort to perform this customization can be significant and needs to be maintained as the application is modified.

    F5 – ASM – F5 offers many modules on their BigIP load balancers, one of which is the ASM. This module adds WAF functionality directly into the load balancer. Additionally, F5 has added policy-based web application security protection.

Software-as-a-service

There are website security software options that include, and sometimes specialize in web scraping protection. This type of solution, from my perspective, is the most effective path.

The SaaS model allows someone else to manage the problem for you and respond with more efficiency even as new threats evolve.  Again, I’m admittedly biased as I co-founded Distil Networks.

When shopping for a SaaS solution to protect against web scraping, you should consider some of the following factors:

•    Does the provider update new threats and rules in real time?

•    How does the solution block suspected non-human visitors?

•    Which types of proactive blocking techniques, such as code injections, does the provider deploy?

•    Which of the reactive techniques, such as rate limiting, are used?

•    Does the solution look at all of your traffic or a snapshot?

•    Can the solution block bots before they reach your infrastructure – and your data?

•    What kind of latency does this solution introduce?

I hope you now have a clearer understanding of web scraping and why it has become so prevalent in travel, and even more important, what you should do to protect and leverage these occurrences.

Source: http://www.tnooz.com/article/what-you-need-to-know-about-web-scraping-how-to-understand-identify-and-sometimes-stop/

Roles of Data Mining in Predicting, Tracking, and Containing the Ebola Outbreak

One of the most diverse continents on earth, Africa astounds the world with its vast savannas and great deserts and with its ancient architecture and modern cities, but Africa also has its share of tragedies and woes.

First identified in Democratic Republic of Congo’s Ebola River in 1976, Ebola Hemorrhagic Fever, a deadly zoonotic disease caused by Ebola virus, has been spreading in West Africa like a wildfire, engulfing everything on its way and creating widespread panic.

What has added insult to injury is the fact that the region has long endured the severe consequences of civil wars and social conflicts, and diseases like malaria, HIV/AIDS, yellow fever, cholera etc. have remained endemic to the region for a long time, causing tens of thousands of deaths every year.

Reportedly, Ebola has already killed at least 2,296 people, and there are about 3,685 confirmed cases of infection. Mortality rate has been swinging between 50% to 90%, depending on the quality of care and nutrition. According to WHO, the disease is likely to infect as much as 20,000 people before it is finally brought under control.

Crisis of Data

When it comes to healthcare management, clinical data is one of the key components. The value of data becomes more urgent in the emergency situation like that of West Africa. The more relevant data you have, the bigger picture you can create for taking aggressive measures. To use Peter Drucker’s words, “What gets measured gets managed.”

Factual data is a precondition for the doctors and health science experts working in the field for measuring and managing the situation. Data helps them to assess their successes or failures and reorient their actions. One of the important reasons why the fight against the Ebola outbreak is turning out into a losing battle is the insufficiency of data. Recently, Scientific American magazine wrote:

Right now, there are not even enough beds for sick patients nor enough data coming in to help track cases. Surveillance and tracking of those who were possibly exposed to Ebola remain inadequate.

In Science magazine, Gretchen Vogel suggests that the death toll of Ebola patients could be much higher than it is currently estimated. She says, “Exactly how many unrecorded Ebola deaths have occurred will never be known. Health officials are keeping track of suspected and probable cases, many of which are people who died before they could be tested.” Greg Slabodkin voices similar concerns in Health Data Management and points at the need of an integrated global biosurveillance system.

The absence of reliable and actionable data has badly hampered the efforts of combatting Ebola and providing proper medical care to the victims. CDC Director Dr. Tom Frieden describes it as a “fog-of-war situation”.

Data Mining: Bots Were the First to Warn

When you flip the coin, however, the situation is not completely bleak and desperate. Even if Big Data technologies have fallen short in predicting, tracking, and containing the epidemic, mainly due to the lack of data from the ground, it has not entirely failed. Data scientists and healthcare experts world over are making concerted efforts to know, track, and defeat the Ebola virus—some on the ground and some in their labs.

The increasing level of collaboration among the biomedical specialists, geneticist, virologists, and IT experts has definitely contributed to slow down the transmission of the virulent disease dubbed as “the plague of modern day”. Médecins Sans Frontières and Healthmap.org are the excellent examples in this regard.

    “By deploying bots and crawlers and by using advanced machine learning algorithms, the Boston-based global infectious disease surveillance system, HealthMap was able to predict and raise concerns about the spread of a mysterious hemorrhagic fever in West Africa nine days earlier than WHO did.”

Run by a team of 45 researchers, epidemiologists, and software developers at Boston Children’s Hospital, HealthMap mines data from search engine queries, social media platforms, health information sites, news reports and crowd-sourced information to track the transmission of the disease and provides an up-to-date timeline report with an interactive map, making it easier for the international health agencies to devise more effective action plans.

HealthMap serves as a good example of how crucial Big Data and data mining technologies could be for handling a healthcare emergency with fact-based and data-driven decisions.

Ebola Data

In their letter to The Lancet, research scientist Rashid Ansumana and his colleagues, working on Ebola in Sierra Leone, highlighted on the need of developing epidemic surveillance systems “by adopting new data-sharing technologies.” They wrote, “Emerging technologies can help early warning systems, outbreak response, and communication between health-care providers, wildlife and veterinary professionals, local and national health authorities, and international health agencies.”

Data-Driven Initiatives to Control the Outbreak

The era of systematic use of data for making better epidemiological predictions and for finding effective healthcare solutions began with Google Flue Trends in 2007, and the rapidly developing tools, technologies, and practices in Big Data have increased the roles of data in healthcare management.

There are a number of data-driven undertakings in progress which have contributed to counter the raging spread of Ebola. Brockmann Lab, run by Professor Dirk Brockmann and his colleagues, for example, has created a computer model for studying correlations and probabilities in the explosion of new cases of infection.

World Airtraffic  Transportation and Relative Import Risk, Source: Brockmann Lab

By applying computational and statistical models, they predict which areas, cities or regions in the world are at the risk of becoming the next Ebola epidemic hotspots. Similarly, Alessandro Vespignani–a network scientist, statistical physicist, and Northeastern professor–has been using human mobility network data to track the cases of Ebola infection and dissemination.

The Swedish NGO Flowminder Foundation has been aggregating, mining, and analyzing anonymized mobile phone location data and is developing national mobility estimates for West Africa to help the local and international agencies to combat the disease.

Meanwhile, innovations with Epi Info VHF, a software tool for case management, contact tracing, analysis and reporting services for Ebola and other hemorrhagic fever outbreaks and OpenStreetMap project for getting location information and spatial data of the affected areas have further helped to guide the intervention initiatives.

However, with all optimism about the growing roles of Big Data and data mining, we also need to be mindful about their limitations. Newsweek aptly puts: “While no media-trawling bot could ever replace national and international health agencies, such tools may be starting to help fill in some of the most gaping holes in real-time knowledge.”

Source: http://www.grepsr.com/blog/data-mining-tracking-ebola-outbreak/

Social Media Crawling & Scraping services for Brand Monitoring

Crawling social media sites for extracting information is a fairly new concept – mainly due to the fact that most of the social media networking sites have cropped up in the last decade or so. But it’s equally (if not more) important to grab this ever-expanding User-Generated-Content (UGC) as this is the data that companies are interested in the most – such as product/service reviews, feedback, complaints, brand monitoring, brand analysis, competitor analysis, overall sentiment towards the brand, and so on.

Scraping social networking sites such as Twitter, Linkedin, Google Plus, Instagram etc. is not an easy task for in-house data acquisition departments of most companies as these sites have complex structures and also restrict the amount and frequency of the data that they let out to crawlers. This kind of a task is best left to an expert, such as PromptCloud’s Social Media Data Acquisition Service – which can take care of your end-to-end requirements and provide you with the desired data in a minimal turnaround time. Most of the popular social networking sites such as Twitter and Facebook let crawlers extract data only through their own API (Application Programming Interface), so as to control the amount of information about their users and their activities.

PromptCloud respects all these restrictions with respect to access to content and frequency of hitting their servers to make sure that user information is not compromised and their experience with the site is unhindered.

Social Media Scraping Experts

At PromptCloud, we have developed an expertise in crawling and scraping social media data in real-time. Such data can be from diverse sources such as – Twitter, Linkedin groups, blogs, news, reviews etc. Popular usage of this data is in brand monitoring, trend watching, sentiment/competitor analysis & customer service, among others.

Our low-latency component can extract data on the basis of specific keywords, categories, geographies, or a combination of these. We can also take care of complexities such as multiple languages as well as tweets and profiles of specific users (based on keywords or geographies). Sample XML data can be accessed through this link – demo.promptcloud.com.

Structured data is delivered via a single REST-based API and every time new content is published, the feed gets updated automatically. We also provide data in any other preferred formats (XML, CSV, XLS etc.).

If you have a social media data acquisition problem that you want to get solved, please do get in touch with us.

Source: https://www.promptcloud.com/social-media-networking-sites-crawling-service/

Web Scraping: Startups, Services & Market

I got recently interested in startups using web scraping in a way or another and since I find the topic very interesting I wanted to share with you some thoughts. [Note that I’m not an expert. To correct me / share your knowledge please use the comment section]

Web scraping is everything but a new technique. However with more and more data shared on internet (from user generated content like social networks & review websites to public/government data and the growing number of online services) the amount of data collected and the use cases possible are increasing at an incredible pace.

We’ve entered the age of “Big Data” and web scraping is one of the sources to feed big data engines with fresh new data, let it be for predictive analytics, competition monitoring or simply to steal data.

From what I could see the startups and services which are using “web scraping” at their core can be divided into three categories:

•    the shovel sellers (a.k.a we sell you the technology to do web scraping)

•    the shovel users (a.k.a we use web scraping to extract gold and sell it to our users)

•    the shovel police (a.k.a the security services which are here to protect website owners from these bots)

The shovel sellers

From a technology point of view efficient web scraping is quite complicated. It exists a number of open source projects (like Beautiful Soup) which enable anyone to get up and running a web scraper by himself. However it’s a whole different story when it has to be the core of your business and that you need not only to maintain your scrapers but also to scale them and to extract smartly the data you need.

This is the reason why more and more services are selling “web scraping” as a service. Their job is to take care about the technical aspects so you can get the data you need without any technical knowledge. Here some examples of such services:

    Grepsr
    Krakio
    import.io
    promptcloud
    80legs
    Proxymesh (funny service: it provides a proxy rotator for web scraping. A shovel seller for shovel seller in a way)
    scrapingHub
    mozanda

The shovel users

It’s the layer above. Web scraping is the technical layer. What is interesting is to make sense of the data you collect. The number of business applications for web scraping is only increasing and some startups are really using it in a truly innovative way to provide a lot of value to their customers.

Basically these startups take care of collecting data then extract the value out of it to sell it to their customers. Here some examples:

Sales intelligence. The scrapers screen marketplaces, competitors, data from public markets, online directories (and more) to find leads. Datanyze, for example, track websites which add or drop javascript tags from your competitors so you can contact them as qualified leads.

Marketing. Web scraping can be used to monitor how your competitors are performing. From reviews they get on marketplaces to press coverage and financial published data you can learn a lot. Concerning marketing there is even a growth hacking class on udemy that teaches you how to leverage scraping for marketing purposes.

Price Intelligence. A very common use case is price monitoring. Whether it’s in the travel, e-commerce or real-estate industry monitoring your competitors’ prices and adjusting yours accordingly is often key. These services not only monitor prices but with their predictive algorithms they can give you advice on where the puck will be. Ex: WisePricer, Pricing Assistant.

Economic intelligence, Finance intelligence etc. with more and more economical, financial and political data available online a new breed of services, which collect and make sense of it, are rising. Ex: connotate.

The shovel police

Web scraping lies in a gray area. Depending on the country or the terms of service of each website, automatically collecting data via robots can be illegal. Whatever the laws say it becomes crucial for some services to try to block these crawlers to protect themselves. The IT security industry has understood it and some startups are starting to tackle this problem. Here are 3 services which claim to provide solutions to stop bots from crawling your website:

•    Distil
•    ScrapeSentry
•    Fireblade

From a market point of view

A couple of points on the market to conclude:

•    It’s hard to assess how big the “web scraping economy” is since it is at the intersection of several big industries (billion dollars): IT security, sales, marketing & finance intelligence. This technique is of course a small component of these industries but is likely to grow in the years to come.

•    A whole underground economy also exists since a lot of web scraping is done through “botnets” (networks of infected computers)

•    It’s a safe bet to say that more and more SaaS (like Datanyze pr Pricing Assistant) will find innovative applications for web scraping. And more and more startups will tackle web scraping from the security point of view.

•    Since these startups are often entering big markets through a niche product / approach (web scraping is a not the solution to everything, there are more a feature) they are likely to be acquired by bigger players (in the security, marketing or sales tools industries). The technological barrier are there.

Source: http://clementvouillon.com/article/web-scraping-startups-services-market/