Network imdb

Network imdb DEFAULT

IMDb TV: Everything You Need to Know About the Fast-Growing Free, Ad-Supported Side of Amazon’s Streaming Business

Teasing out the raison d’être and market role of IMDb TV, the free service tucked into Amazon’s panoply of video offerings, can be a challenge for all but the most closely attuned observer. 

It’s not another Amazon Prime Video, which features pricey AAA originals as a particularly attractive benefit of the $119 annual Prime subscription. It’s definitely not Twitch, the live-streaming service moving away from its gaming roots. And it’s not Channels, Amazon’s add-on service selling subscriptions to about 150 third-party channels, including a recently expanded its partnership with AMC Networks to launch five live-linear channels.  

Also Read: AVOD Platforms Suffer From Low Consumer Awareness: Magid

More than three years after launch, IMDb TV is changing up its original strategy moving from mostly “shoulder content” for fans of film and TV to more high profile fare.

In October, it was announced that Judge Judy Sheindlin would get a show on IMDb TV. “The Untitled Judge Judy Sheindlin Project will feature no-nonsense, expeditious Judge Sheindlin’s signature adjudication style with a new array of cases, litigants and judgments,” Amazon said.

November saw the premiere of Alex Rider, based on a series of novels by Anthony Horowitz. IMDb TV renewed the spy series ahead of its season one premiere on the service.

Amazon is also working on a Bosch spinoff series for IMDb TV in which Titus Welliver will reprise the title role.

On the non-fiction side, IMDb TV's expanded relationship with corporate sibling Amazon Studios will see the AVOD service get original docu-series Moment of Truth, about the murder of Michael Jordan's father, and an untitled series on the cultural impact of TV crime procedurals.     

The service offers an impressive bench of licensed shows as well, culled from the libraries of Fox, Universal, DreamWorks, MGM, Sony, Paramount and others. Among the available movies are such notables as Patriot Games, O Brother Where Art Thou, Taxi Driver, Oldboy, Sherlock Holmes, Rushmore, Life of Pi, Lawrence of Arabia, Charade, Whale Rider, His Gal Friday, Spotlight and Memento. 

IMDb TV also got in on the action when Amazon signed an exclusive "Pay 1" distribution deal with Universal Pictures recently. The AVOD platform will soon get streaming rights to Universal's 2021-21 film slate, which includes the latest, incomprehensibly popular The Fast and the Furious movie. 

Meanwhile, available TV series include: Ugly Betty, Malcolm in the Middle, Boston Legal, Murder She Wrote, St. Elsewhere, Little House on the Prairie, Ally McBeal, Desperate Housewives, Columbo, The Rockford Files, Wiseguy, Dallas, The Rifleman, and Mad Men,part of a complicated rights deal Lionsgate cut with Amazon and several other outlets after a long run on Netflix.

Other acquired series for IMDb TV since its content team moved under the Amazon Studios umbrella include Chicago Fire, Schitt’s Creek, Lost, and Friday Night Lights. And it’s all free to watch, for the price of registering.

If you do subscribe to Prime, IMDb TV becomes one of your available Channels. As the site proclaims in a 5-second ad before every show, it’s “Always Entertaining. Always Free.”

Perhaps most notably, Amazon revealed in February that it has around 55 million active users for its ad-supported content.

The ad support is lighter than typical for broadcast or pay-TV outlets, more like a Hulu. As the promos also say, “Fewer ads, No Subscriptions required.”

In O Brother, for instance, the 107-minute film had seven ad “pods,” each with two or three brief ads totaling about a minute of run time. Not bad when watching a 20-year-old Coen Bros. movie that grabbed a couple of Oscar nominations, won George Clooney a Golden Globe, and featured one of the best-selling movie soundtracks of all time. 

More Than a Database

By the way, I know all that because of one of IMDb TV’s defining aspects: its integration with the authoritative database of film and TV shows that Amazon bought in 1998, basically as a lead-generation tool for the DVDs and VHS tapes it was selling back then. 

IMDb TV is much newer, launching in January 2019 as FreeDive, a puzzling name that had no connection to any Amazon asset, nor any resonance with consumers. It took Amazon about six months to rename the service to one of its oldest and most prominent in-house brands. Since then, the service seems focused as much on providing TV and movie fans stuff about TV and films as it is about actually streaming those shows to them. 

IMDb TV, in fact, isn’t even the first thing you see on the mobile app. These days, the first screen spotlights fan-focused meta-content, then (if you’re signed in), your personalized watchlist of upcoming programs, and prominent programming on major streamers. 

“Fan Favorites” include the week's top TV and movies, featuring offerings from Disney Plus, Netflix, HBO and other outlets. WandaVision, Dark and The Old Guard on Netflix, Hamilton on Disney+, and Joker on HBO.

Where Does IMDb TV Land in the Amazon Firmament?

Analyst Rich Greenfield of Lightshed Partners suggests that IMDb TV ultimately should be seen as part of an array of Amazon services designed to keep customers engaged within its ecosystem. Amazon, like Apple and Google, is playing a much broader strategic game, offering customers a spectrum of video, music and gaming entertainment, and the devices to deliver them. 

Add the various entertainment offerings from all three tech giants to their efforts in smart-home technology, immersive entertainment, AI-based personal assistants and speakers, and a raft of other devices from iPhones to Fire TVs to Nest thermostats. IMDb TV is just another soldier in the battle to keep consumers engaged with Amazon for nearly everything. 

Perhaps that’s why tracking down IMDb TV can be a bit of challenge. It’s the third option on the IMDb app, doesn’t have its own separate app, isn’t available on Apple TV, and is tucked deep in the Amazon Prime website. It is, of course, closely integrated into Amazon Fire TV devices.

The AVOD offering did recently expand distribution to LG smart TVs, PlayStation 4, Nvidia Shield and TiVo Stream 4K. IMDb TV also has Google TV app support and has been added to Amazon Fire TV's live lineup.

And there’s plenty to watch, from the high-profile licensed programs to lots of original material about beloved stars, shows, genres and their back stories. Keeping those hard-core fans around, and giving them more reasons to watch, and subscribe to third-party channels, and buy seasons of favorites, all while driving ad revenue to pay for it all, may be all the reason to be that IMDb TV needs in Amazon World. 


TLDR; Short on time? IMDbPro is my #1 recommendation if you’re looking for a tool to build your network of creative professionals in the entertainment industry. The ten features I detail in this article are just a small portion of what it has to offer and in my opinion make it worth the price of admission (and then some). IMDbPro isn’t just for looking up film credits!

If 2020 has taught us anything, it’s that you can’t rely on your old, comfortable way of doing things anymore. And nowhere is this more obvious than trying to build your professional network and find gigs during a pandemic.

Love networking? Do you enjoy going to mixers, in-person meetups, and panel events to rub elbows with as many people as possible, ultimately passing around your contact info in hopes of getting a job lead? Too bad, no in-person events for you this year!


Are you an introvert who hates going to live events? Do you rely on past and existing colleagues to let you know what’s available and refer you for the next gig? Nope! Wrong again. There hasn’t been enough work to go around for anyone as it is this year. Find your own jobs.

Hate networking in all of its forms? Do you prefer to hunker down, do amazing work in the comfort of your windowless room, perfect your creative craft, and wait around (with your white-knuckled fingers desperately crossed) hoping that someone eventually “discovers you?” Pandemic or no…how’s that working out?

No matter what your networking strategy may have been pre-pandemic to build relationships and find your next gig, virtual networking is the only game in town for the foreseeable future. Which means that if you need to find work, or conversely if you want to maximize this time to make a major transition in your career, you have to find a way to do it from home.

Don’t worry…as a self-proclaimed extreme introvert who’s been social distancing at the Olympic level since 2005…I’ve got your back. 😎

Yes, you could continue with the ‘Shotgun Approach,’ sending yet another cringe-worthy group email telling all of your past colleagues “I’m available! If you hear of anything please pass along my resume (attached). Thanks!” (P.S. I don’t recommend this strategy).

But what if instead you became a ‘Networking Sniper’ and used the time & resources that are available to you right now to build your ‘Dream Network’ of colleagues that could provide you with much needed career advice, offer you a shadowing opportunity (yes, virtual shadowing is a thing!), become your mentor, or even refer you for an upcoming gig?

Luckily there’s a single tool to help you build your professional network in the entertainment industry that’s just as indispensable to your career as Avid, Premiere, After Effects, FCP X, Final Draft, Resolve, Pro Tools, or otherwise.

That indispensable tool, in my humble opinion, is IMDbPro. (P.S. This is not an ad*)

Having used IMDbPro as my #1 tool for research and virtual networking for over 15 years now, my intention by the end of this article is not only to convince you that IMDbPro is worth the price of admission but also demonstrate the many features IMDbPro offers that help you present yourself as professionally as possible while simultaneously doing the best networking of your career…from the comfort of your pajamas.

*Author’s note: I am in no way affiliated with IMDbPro, nor am I receiving any compensation for recommending their product. In the interest of full disclosure, I have been granted free membership for offering training seminars to industry professionals teaching them how to specifically use the networking features of IMDbPro.

What’s the Difference Between IMDb and IMDbPro?

If you weren’t even aware that IMDb offered a Pro version and you’re wondering what the difference is, it’s pretty simple:

IMDb is an entertainment website for fans.

IMDbPro is the equivalent of LinkedIn for entertainment professionals.

No offense to the people who designed IMDb, but it’s one flashing banner ad away from requiring an epileptic seizure warning. It’s a mess of advertising from top to bottom, pop-up videos, auto scrolling images, and endless recommendations, lists, and news feeds.

imdb home page screenshot

The IMDb Home Page
Call me 100 years old, but IMDb is an annotated scribble away from Perez Hilton.

IMDbPro, on the other hand, contains the exact same database of information as IMDb (plus access to a whole lot more), but luckily with your paid subscription they have done away with all the noise and clutter to make it a pleasurable research experience devoid of the vast majority of noise and distractions that could keep you from the valuable networking information that is at your fingertips.

imdbpro home page

The IMDbPro Home Page
An industry dashboard built for industry professionals with news, box office info, networking features, project tracking info, and more right from page one.

If you’re ready to take the plunge and give the IMDbPro Trial Subscription a whirl (it’s free for 30 days…and I’m serious, this is NOT an ad), or if you already have a subscription but you feel you’re not getting your money’s worth, below I have provided 10 killer features that will help you get the most out of IMDbPro so you can identify & organize all of the projects and people that belong in your ‘Dream Network.’

Five IMDbPro Features That Make It Worth the Price of Admission

Below I have outlined five simple features that IMDbPro offers you that take you well beyond the basic functionality of the standard site. (For the ninja-level networking features, scroll below to the next section.)

1. Access to direct contact info for a specific person, agents & managers, and even for an entire production company (none of which is visible in IMDb)

When it comes to cold outreach, one of the most common objections I hear from students in my coaching & mentorship community is:

“How can I connect with people if I can’t find their contact info?”

With an IMDbPro subscription you get insider access to contact info for individuals, their team (e.g. agents, managers, publicists, etc), and if you want to connect with a production company because they produce projects you’re interested in pursuing, IMDbPro provides that contact info too (assuming the individual or company provided it).


Steven Spielberg’s IMDb page vs. IMDbPro with TONS of contact info
(If you reach out, tell Steven that Zack says hi!)

2. Customization options to update your primary profile photo and upload multiple additional photos (BONUS: You can upload your reel too!)

If you are an actor, IMDbPro is just as important as your headshot and resume (possibly even more so). But more than likely you are crew, so do photos even matter? YES!!!! First impressions mean everything, especially in Hollywood. The first image someone sees of you will help them better understand if you are the kind of person they’d like to work with.

Are you smiling?

Do you come off as warm and friendly?

Do you seem like someone they’d love to work with for hours on end?

A picture is worth a thousand words, and IMDbPro gives you the ability to not only upload multiple images but also choose which will become your ‘featured image’ to sell you for your next job.

zack arnold imdbpro pageMy IMDbPro profile with a featured image (of my choosing) plus additional images

3. The ability to customize your ‘Known For’ featured credits

Even when I was editing Burn Notice for four straight seasons I still needed to find work during the gaps. And having control over the ‘Known For’ feature made the monthly price worth it all by itself to help me land the next in-between gig. Whether we want to accept it or not, credits matter in Hollywood. What you have worked on (especially lately) has a huge impact on your next job.

Do you want a short film that you were an assistant on 12 years ago to be in your ‘Known For’ section just because the IMDb algorithm says so?

Or instead would you prefer to have control over the story you tell about your past jobs that will best suit the direction your career is headed next?

Don’t let IMDb pigeonhole you. Take control of your profile and make it clear which projects you want to highlight to prospective employers.

imdbpro page known for sectionI’ve paid the monthly fee just so I could keep this section of my profile updated

4. The ability to customize what profile info is public vs private

Depending on the specific job you want to apply for or the person you want to connect with, there might be certain information on your profile that you’d like to hide temporarily. Rather than submitting a change and potentially waiting days for your profile to update, you have the ability to customize what profile info is public with a simple Yes/No switch.

Customize your profile as needed depending on who you’re contacting next

This might seem like a silly feature, but this one option can help improve the chances of those who might be concerned about ageism in the hiring process (Yes…ageism is a real thing in Hollywood) because you can even hide your birthday if necessary (note that it will still appear on the regular IMDb website…don’t shoot the messenger).

The Display/Hide feature can also be tremendously useful if you want to display your direct contact info just for a short period of time while being considered for a specific opportunity but have it hidden the rest of the time.

5. Access to customer support to help you fix all your incorrect (or missing) credits

There are few things more infuriating than that moment someone asks for a link to your IMDb page when you know there’s a bunch of incorrect information. It’s a user database, after all, so I guess it just is what it is, right? With the paid subscription it doesn’t have to be!

Is your profile missing credits?

Were you credited improperly for something? Or uncredited?

Do you have multiple accounts that need to be merged?

Need to unclaim a page that was incorrectly attributed to you?

Yes, you could continue to submit via IMDb Wikipedia-style (technically IMDb is a lot better vetted than Wikipedia…but we all know mistakes still fall through the cracks). As an IMDbPro member you have access to 24/7 support directly from the IMDbPro customer service team. If you have a last-minute job interview, the wrong credits on your IMDb page could make or break your chances.

Getting a priority response from support could be the difference between success and failure (I bet Alan Smithee wished he had access to IMDbPro support).

Five Ninja-Level Networking Features In IMDbPro That Will Help You Build Your ‘Dream Network’

Alright…now that you’ve taken the steps to get your profile updated so it tells the right story about your past work history, let’s talk about all the features that are totally MONEY if you want to take IMDbPro to the next level so you can virtually build your dream network from your laptop.

Ninja Feature #1: Searchable production listings so you can identify and sort the best opportunities for your needs.

The only time I’m working harder than when I’m on a job…is when I’m looking for a job. And if I’m going to put in that level of effort to find my next gig, I want to be the most effective with my time as possible. I’m a big believer that unless you have absolutely no other choice financially, you should never simply take the next gig at the expense of building your career. I play chess with my resume instead of checkers. And to ensure I have access to all the latest production information so I can get my name into consideration before everything is crewed up, I have often relied on IMDbPro’s ‘Production Listings’ feature.

With Production Listings you can search by the following criteria:

  • Type (Movie, TV Series, TV Movie, etc)
  • Status (In Development, Pre-Production, etc)
  • MOVIEmeter
  • Release Year
  • Location
  • Budget
  • US Gross
  • Country
  • Genre

If you’ve identified that the next best step in your career is working on a season of TV you’re already a fan of (i.e. at least one season has specifically been released), and it’s a comedy, this is what your list might look like:

imdbpro-page-production-listingAn example of comedy TV Series in:

Ninja Feature #2: Project & people tracking so you get notifications about any relevant topics.

Once you have identified projects that are potentially a good fit for you, the next step is organizing and tracking those projects and identifying the people involved so you can get your foot in the door, make a good impression, and potentially work on that show!

The IMDbPro “Track” feature is great for managing the multiple people & projects that belong on your ‘Dream List.’ More importantly, with projects actively in production you’ll receive periodic notifications for all news or status updates pertaining to anything on your list.

imdbpro-track-featureUse the ‘Track’ feature to make sure you never miss updates about your favorite projects & people (assuming you have notifications enabled)

Ninja Feature #3: The ability to create ‘Lists’ to keep your people & projects organized

Having gone through the Production Listings and chosen projects and/or people you want to track, things will get disorganized very quickly if you don’t have lists (if you didn’t know already, I’m pretty OCD when it comes to organizing data).

The ‘Add to List’ feature allows you to build as many lists as you’d like with different projects and people based on the unique criteria you choose, for example you could collect all of the dream projects that you’d love to edit (including if you had a time machine and could have done so in the past).

imdbpro-dream-projectsMy five dream editing jobs!

BONUS LEVEL: Once you have identified your dream projects, you can even go one step further and begin to organize all of the people that worked on those projects in the key roles that apply to your goals. As you begin to sort and organize names across different shows you’ll begin to recognize patterns and repeated names – Hollywood is a small town. These repeated names that keep jumping off the page are the ‘sweet spot’ where you should most likely focus your networking attention next.

I will admit this sorting, organizing, and prioritizing process is an area where IMDbPro isn’t as strong as it might be with its other features. Therefore I created my own spreadsheet to help me both organize and prioritize all the data I collect from IMDbPro so I can always identify (even in a list of hundreds of names) who I need to build a relationship with NEXT.

The following is a brief excerpt of an introductory lesson from my new Masterclass:
Virtually Build Your Network Using IMDbPro

» Click here to get your own working copy of this spreadsheet template

Ninja Feature #4: The ability to filter credits by specific professions (this is a HUGE time saver)

If you’ve used the Production Listings feature for tracking and creating lists, inevitably you’ll need to begin prioritizing the people you need to contact in order to open the right doors for the next stage in your career. And with IMDb that could get messy – FAST.

With IMDbPro you have the option to ‘Filter by profession’ to eliminate the noise and collect information for only the people you need to connect with. This feature alone is a HUGE time saver for gathering contact data.


imdbpro-filter-by-professionThe ‘Filter by profession’ option is a huge time-saver for contact organization

Ninja Feature #5: The ‘Networking’ tab to help you identify mutual connections (and facilitate warm introductions)

Now that you’ve dug into the many features shared above that allow you to transform IMDbPro from an entertainment resource to the number one networking tool in your arsenal, here’s the feature that will CHANGE THE GAME – the ‘Network’ tab.

The ‘Network’ tab is as close as you’re going to get to using IMDbPro as you would LinkedIn. I’ll admit the feature is far from perfect in its current iteration, but it’s helped me facilitate many mutual introductions that led to relationships I might not have been able to build had I reached out cold.

For example, let’s say I want to connect with legendary editor Walter Murch to seek advice or potential mentorship. I could do something crazy like write him a physical letter, send it to his house, and hope he responds (which btw I did years ago…and he did respond!). But if you’re not as bold as I might have been in my earlier years, you could instead use the ‘Network’ tab to identify a mutual introduction.

After using this feature I discovered (in a single click) a mutual colleague that both Walter Murch and I had worked with in the past (they were SECOND on this list). While this doesn’t guarantee you’ll discover a strong mutual connection (this feature currently yields a lot of very random connections), every once in a while you find a diamond in the rough that leads to an amazing mutual introduction.

imdbpro-network-feature-walter-murchUsing the ‘Network’ feature I found a direct connection to Walter Murch via a DP I worked with on an indie feature several years ago

IMDbPro is THE Essential Tool For Career-Building In the Entertainment Industry

I know we creatives love our shiny objects like Avid, After Effects, Premiere, Resolve, and so many more, but in my opinion IMDbPro is an equally essential tool if you intend to build a lasting and fulfilling career path (as opposed to constantly chasing after the next gig).

IMDb is no longer for simply checking to see who did what on whatever show while you’re binging Netflix. It’s bigger brother IMDbPro has a robust toolset that allows you to precisely identify, organize, and prioritize the most important people and projects that can take your career to a whole new level, including:

  • Access to direct contact info
  • Profile customization options
  • Customer support for last-minute fixes and upgrades before that important meeting
  • Searchable production listings
  • Project & people tracking
  • The ability to sort with Lists
  • Filtering credits by profession
  • The ‘Network’ tab to find mutual connections

You could take the ‘Shotgun approach’ and reach out to everyone you know (and even those you don’t), blanketing the market with group emails and your attached resume hoping someone bites…

…OR you could take the ‘Sniper approach’ and identify exactly the right projects that you would find fulfilling and the people who can open doors to the next stage in your career.

Ready to take your networking skills to a whole new level?

*Author’s note: I am in no way affiliated with IMDbPro, nor am I receiving any compensation for recommending their product. In the interest of full disclosure, I have been granted free membership for offering training seminars to industry professionals teaching them how to specifically use the networking features of IMDbPro.

Continue to Listen & Learn:

Why Writing Cold Emails Is the Most Important ‘Soft Skill’ You Must Master (Especially If You’re An Introvert)

The 5 Most Common Mistakes You’re Making With Your Cold Emails (and How to Fix Them)

Want a Response to Your Cold Emails? Then Every Message MUST Do These 5 Things

Learn about my ‘Optimizer’ Coaching & Mentorship Community

How I Landed My Dream Job Editing ‘Cobra Kai’ | with Jon Hurwitz, Josh Heald, and Hayden Schlossberg

Get your own working copy of the IMDbPro Networking spreadsheet template →

Legendary Editor Walter Murch On Optimizing Creativity, Productivity, and Well-Being In Hollywood For 50+ Years

Chasing After the Next Gig vs. Building Your Career | with Kabir Ahktar, ACE

On the Importance of Building Relationships, Asking Questions, and Never Giving Up | with Andi Armaganian

How to Create Your Own Opportunities to Pursue Dream Projects | with Richard Sanchez

Zack Arnold

Zack Arnold (ACE) is an award-winning Hollywood film editor & producer (Cobra Kai, Empire, Burn Notice, Unsolved, Glee), a documentary director, father of 2, an American Ninja Warrior, and the creator of Optimize Yourself. He believes we all deserve to love what we do for a living...but not at the expense of our health, our relationships, or our sanity. He provides the education, motivation, and inspiration to help ambitious creative professionals DO better and BE better. “Doing” better means learning how to more effectively manage your time and creative energy so you can produce higher quality work in less time. “Being” better means doing all of the above while still prioritizing the most important people and passions in your life…all without burning out in the process. Click to download Zack’s “Ultimate Guide to Optimizing Your Creativity (And Avoiding Burnout).”

  1. Blackout mini h frame
  2. Taco tuesday posters
  3. Healthy lifestyle icons
  4. Dragon disciple


For the database system, see In-memory database. For the Malaysian sovereign wealth fund, see 1MDB.

Online database for movies, television, and video games

IMDB Logo 2016.svg


IMDb homepage.png

IMDb homepage as of March 2021

Type of site

Online database for movies, television, and video games
Available inEnglish
SubsidiariesBox Office Mojo
RegistrationOptional registration; registered members can write reviews, edit the site, vote on ratings
LaunchedOctober 17, 1990; 31 years ago (1990-10-17)
Current statusActive

IMDb (an acronym for Internet Movie Database)[1] is an online database of information related to films, television programs, home videos, video games, and streaming content online – including cast, production crew and personal biographies, plot summaries, trivia, ratings, and fan and critical reviews. An additional fan feature, message boards, was abandoned in February 2017. Originally a fan-operated website, the database is now owned and operated by, Inc., a subsidiary of Amazon.

As of June 2021[update], IMDb has approximately 8 million titles (including episodes) and 10.4 million personalities in its database,[2] as well as 83 million registered users.

IMDb began as a movie database on the Usenet group "rec.arts.movies" in 1990 and moved to the web in 1993.


The movie and talent pages of IMDb are accessible to all internet users, but a registration process is necessary to contribute information to the site.

Most data in the database is provided by volunteer contributors. The site enables registered users to submit new material and edits to existing entries. Users with a proven track record of submitting data are given instant approval for additions or corrections to cast, credits, and other demographics of media product and personalities. However, image, name, character name, plot summaries, and title changes are supposedly screened before publication, and usually take between 24 and 72 hours to appear.

All registered users can choose their own site name, and most operate anonymously. They have a profile page which shows how long a registered user has been a member, as well as personal movie ratings (should the user decide to display them) and, since 2015, "badges" are added representing how many contributions a particular registered user has submitted. These badges range from total contributions made to independent categories such as photos, trivia, bios, etc. If a registered user or visitor is in the entertainment industry and has an IMDb page, then that user/visitor can add photos to that page by enrolling in IMDbPRO.[3] There is no single index of contributors, no index on each profile page of the items contributed, and (except for plot synopses and biographies) no identification of contributors to each product's or person's data pages.

Users are also invited to rate any film or TV series on a scale of 1 to 10, and the totals are converted into a weighted mean-rating that is displayed beside each title, with online filters employed to deter ballot-stuffing.

In January 2019, IMDb launched a free movie streaming platform called Freedive, an ad-supported service offering Hollywood movie titles and TV shows. Many Freedive titles are licensed from Sony Pictures.[4] Subsequently, in June 2019, Freedive was rebranded as IMDbTV, during the launch of which, the amount of content contained on the platform was tripled.[5]


History before website[edit]

IMDb originated in 1990 with a Usenet posting entitled "Those Eyes", by the British film fan and computer programmer Col Needham,[6] about actresses with beautiful eyes. Others with similar interests soon responded with additions or different lists of their own. Needham subsequently started an "Actors List", while Dave Knight began a "Directors List", and Andy Krieg took over "THE LIST" from Hank Driskill, which would later be renamed the "Actress List". Both lists had been restricted to people who were alive and working, but soon retired people were added, so Needham started what was then (but did not remain) a separate "Dead Actors/Actresses List". Steve Hammond started collecting and merging character names for both the actors and actresses lists. When these achieved popularity, they were merged back into the lists themselves. The goal of the participants now was to make the lists as inclusive as possible.

By late 1990, the lists included almost 10,000 films and television series, correlated with actors and actresses appearing therein. On October 17, 1990, Needham developed and posted a collection of Unixshell scripts that could be used to search the four lists, and thus the database that would become the IMDb was born.[7] At the time, it was known as the "rec.arts.movies movie database".

On the web[edit]

The database had been expanded to include additional categories of filmmakers and other demographic material as well as trivia, biographies, and plot summaries. The movie ratings had been properly integrated with the list data, and a centralized email interface for querying the database had been created by Alan Jay. Later, in 1993, it moved onto the fledgling World Wide Web under the name of Cardiff Internet Movie Database.[8] The database resided on the servers of the computer science department of Cardiff University in Wales. Rob Hartill was the original web interface author. In 1994, the email interface was revised to accept the submission of all information, which enabled people to email the specific list maintainer with their updates. However, the structure remained so that information received on a single film was divided among multiple section managers, the sections being defined and determined by categories of film personnel and the individual filmographies contained therein. Over the next few years, the database was run on a network of mirrors across the world with donated bandwidth.[9]

As an independent company[edit]

In 1996 IMDb was incorporated in the United Kingdom, becoming the Internet Movie Database Ltd. Founder Col Needham became the primary owner. General revenue for site operations was generated through advertising, licensing and partnerships.[10]

As subsidiary (1998–present)[edit]

In 1998, Jeff Bezos, founder, owner, and CEO of, struck a deal with Needham and other principal shareholders to buy IMDb outright; Amazon paid $55 million for IMDb and two other companies.[11] Bezos attached it to Amazon as a subsidiary, private company.[12] This gave IMDb the ability to pay the shareholders salaries for their work. In the process of expanding its product line, intended to use IMDb as an advertising resource for selling DVDs and videotapes.

IMDb continued to expand its functionality. On January 15, 2002, it added a subscription service known as IMDbPro, aimed at entertainment professionals. IMDbPro was announced and launched at the 2002 Sundance Film Festival. It provides a variety of services including film production and box office details, a company directory, and the ability of subscribers to add personal information pages with details at variance.

As an additional incentive for users, as of 2003[update], users identified as one of "the top 100 contributors" of hard data received complimentary free access to IMDbPro for the following calendar year; for 2006 this was increased to the top 150 contributors, and for 2010 to the top 250.[13] In 2008, IMDb launched their first official foreign-language version with the German Also in 2008, IMDb acquired two other companies: Withoutabox[14] and Box Office Mojo.[15]

The website was originally Perl-based, but IMDb no longer discloses what software it uses for reasons of security.[16] In 2010, the site was filtered in China.[17]

In 2016, The IMDb Studio at Sundance was launched, a talk show that is presented on IMDb and YouTube.[18][19]

In April 2017, IMDb celebrated its 25th anniversary. As of that year, Needham was still managing IMDb from its main office in Bristol in the Castlemead office tower.[20]


Actors, crew, and industry executives can post their own resume and upload photos of themselves for a yearly membership fee to IMDbPro. IMDbPro can be accessed by anyone willing to pay the fee, which is US$19.99 per month, or if paid annually, US$149.99. Membership enables a user to access the rank order of each industry personality, as well as agent contact information for any actor, producer, director etc. that has an IMDb page. IMDbPro also allows existing actors to claim their name page. Enrolling in IMDbPro enables members who are industry personnel to upload a head shot to open their page, as well as to upload hundreds of photos to accompany their page. Anyone can register as an IMDb user and contribute to the site as well as view its content; however, those users enrolled in IMDbPro have greater access and privileges.[21]

Characters' filmography[edit]

On October 2, 2007,[citation needed] the characters' filmography was added. Character entries are created from character listings in the main filmography database, and as such do not need any additional verification by IMDb staff. They have already been verified when they are added to the main filmography.

IMDb TV[edit]

Main article: IMDb TV

On January 10, 2019, IMDb launched Freedive, a streaming service that offers certain films free with advertisements. Due to international licensing restrictions, the service currently is only available in the United States as of 2019[update].[22]

Freedive was the company's second attempt at a streaming service. IMDb had previously launched a similar feature in 2008[23][24] that was later discontinued.[citation needed]

In June 2019, IMDb announced that its streaming service would be renamed IMDb TV, with a planned launch in Europe later that year. The service has licensed films and TV series from Warner Bros., Sony Pictures Entertainment and MGM.[25] In February 2020, the service licensed 21 television series from Disney–ABC Domestic Television.[26]

IMDb TV has also bought the streaming rights to the Alex Rider TV series in the U.S that has been streaming on Amazon Prime Video in the UK, Latin America, Germany and Austria.[27]

Content and format[edit]

Data provided by subjects[edit]

In 2006, IMDb introduced its "Résumé Subscription Service", where actors and crew can post their own résumé and upload photos of themselves[28] for a yearly fee.[29] The base annual charge for including a photo with an account was US$39.95 until 2010, when it was increased to US$54.95. IMDb résumé pages are kept on a sub-page of the regular entry about that person, with a regular entry automatically created for each résumé subscriber who does not already have one.[30]

As of 2012[update], Resume Services is now included as part of an IMDbPro subscription and is no longer offered as a separate subscription service.

Copyright, vandalism and error issues[edit]

All volunteers who contribute content to the database technically retain copyright on their contributions, but the compilation of the content becomes the exclusive property of IMDb with the full right to copy, modify, and sublicense it, and they are verified before posting.[31] Credit is not given on specific title or filmography pages to the contributor(s) who have provided information. Conversely, a credited text entry, such as a plot summary, may be corrected for content, grammar, sentence structure, perceived omission or error, by other contributors without having to add their names as co-authors. Due to the time required for processing submitted data or text before it is displayed, IMDb is different from user-contributed projects like Discogs, or OpenStreetMap, or Wikipedia, in that contributors cannot add, delete, or modify the data or text on impulse, and the manipulation of data is controlled by IMDb technology and salaried staff.[32]

IMDb has been subject to deliberate additions of false information; in 2012 a spokesperson said: "We make it easy for users and professionals to update much of our content, which is why we have an 'edit page.' The data that is submitted goes through a series of consistency checks before it goes live. Given the sheer volume of the information, occasional mistakes are inevitable, and, when reported, they are promptly fixed. We always welcome corrections."[33]

The Java Movie Database (JMDB)[34] is reportedly creating an IMDb_Error.log file that lists all the errors found while processing the IMDb plain text files. A Wiki alternative to IMDb is Open Media Database[35] whose content is also contributed by users but licensed under CC-by and the GFDL. Since 2007, IMDb has been experimenting with wiki-programmed sections for complete film synopses, parental guides, and FAQs about titles as determined by (and answered by) individual contributors.

Data format and access[edit]

IMDb does not provide an API for automated queries. However, most of the data can be downloaded as compressed plain text files and the information can be extracted using the command-line interface tools provided.[36] There is also a Java-based graphical user interface (GUI) application available that is able to process the compressed plain text files, which allows a search and a display of the information.[34] This GUI application supports different languages, but the movie related data are in English, as made available by IMDb. A Python package called IMDbPY can also be used to process the compressed plain text files into a number of different SQL databases, enabling easier access to the entire dataset for searching or data mining.[37]

Film titles[edit]

The IMDb has sites in English as well as versions translated completely or in part into other languages (Danish, Finnish, French, German, Hungarian, Italian, Polish, Portuguese, Spanish and Romanian). The non-English language sites display film titles in the specified language. Originally, IMDb's English language sites displayed titles according to their original country-of-origin language, however, in 2010 IMDb began allowing individual users in the UK and USA to choose primary title display by either the original-language titles, or the US or UK release title (normally, in English).

Ancillary features[edit]

User ratings of films[edit]

As one adjunct to data, the IMDb offers a rating scale that allows users to rate films on a scale of one to ten.

IMDb indicates that submitted ratings are filtered and weighted in various ways to produce a weighted mean that is displayed for each film, series, and so on. It states that filters are used to avoid ballot stuffing; the method is not described in detail to avoid attempts to circumvent it. In fact, it sometimes produces an extreme difference between the weighted average and the arithmetic mean.


The IMDb Top 250 is a list of the top rated 250 films, based on ratings by the registered users of the website using the methods described. As of 8 June 2021[update], The Shawshank Redemption is No. 1 on the list.[38] The "Top 250" rating is based on only the ratings of "regular voters". The number of votes a registered user would have to make to be considered as a user who votes regularly has been kept secret. IMDb has stated that to maintain the effectiveness of the Top 250 list they "deliberately do not disclose the criteria used for a person to be counted as a regular voter".[39] In addition to other weightings, the Top 250 films are also based on a weighted rating formula referred to in actuarial science as a credibility formula.[40] This label arises because a statistic is taken to be more credible the greater the number of individual pieces of information; in this case from eligible users who submit ratings. Although the current formula is not disclosed, IMDb originally used the following formula to calculate their weighted rating:[41]

{\displaystyle W={\frac {\ R\cdot v+C\cdot m\ }{v+m}}}

The W\ in this formula is equivalent to a Bayesian posterior mean (see Bayesian statistics).

The IMDb also has a Bottom 100 feature which is assembled through a similar process although only 10,000 votes must be received to qualify for the list.[42]

The Top 250 list comprises a wide range of feature films, including major releases, cult films, independent films, critically acclaimed films, silent films, and non-English-language films. Documentaries, short films and TV episodes are not currently included.

Since 2015, there has been a Top 250 list devoted to ranking television shows.[43]

Message boards[edit]

Beginning in 2001, the Internet Movie Database also maintained message boards for every title (excepting, as of 2013[update], TV episodes[44]) and name entry, along with over 140 main boards. To post on the message boards a user needed to "authenticate" their account via cell phone, credit card, or by having been a recent customer of the parent company Message boards expanded in recent years. The Soapbox started in 1999 as a general message board meant for debates on any subjects. The Politics board started in 2007 was a message board to discuss politics, news events, and current affairs, as well as history and economics.[citation needed]

By February 20, 2017, all the message boards and their content were permanently removed. According to the website, the decision was made because the boards were "no longer providing a positive, useful experience for the vast majority of our more than 250 million monthly users worldwide",[45] and others have mentioned its susceptibility to trolling and disagreeable behavior.[46][47][48] Col Needham also mentioned in a post some months earlier that the boards received less income from ads, and that their members only made up a very small part of the website's visitors. The boards were costly to run due to the system's age and dated design, which did not make business sense.[49] The decision to remove the message boards was met with outspoken backlash from some of its users, and sparked an online petition garnering over 8,000 signatures.[50] In the days leading up to February 20, 2017, both[51] and[52] preserved the entire contents of the IMDb message boards using web scraping. and have published IMDb message board archives, which is legal under the fair use doctrine, because it has no effect on IMDb's potential market for or value of the copyrighted work.[53][54]


As of June 2021[update], IMDb has the following statistics:[2]

Feature film 580,890
Short film 818,520
TV series 209,880
TV episode 5,884,536
TV movie 132,040
TV special 33,183
TV mini-series 38,108
TV short 9,949
Video 312,751
Video game 28,467


In 2011, in the case of Hoang v., Inc., IMDb was sued by an anonymous actress for more than US$1,000,000 due to IMDb's revealing her age (40, at the time).[55] The actress claimed that revealing her age could cause her to lose acting opportunities.[56] Judge Marsha J. Pechman, a U.S. district judge in Seattle, dismissed the lawsuit, saying the actress had no grounds to proceed with an anonymous complaint. The actress re-filed and so revealed that she was Huong Hoang of Texas, who uses the stage name Junie Hoang.[57] In 2013, Pechman dismissed all causes of action except for a breach of contract claim against IMDb; a jury then sided with IMDb on that claim.[58] The Court of Appeals for the Ninth Circuit affirmed the district court judgment in March 2015.[59]

Also in 2011, in the case of United Video Properties Inc., et al. v. Amazon.Com Inc. et al.,[60] IMDb and Amazon were sued by Rovi Corporation and others for patent infringement over their various program listing offerings.[61] The patent claims were ultimately construed in a way favorable to IMDb, and Rovi / United Video Properties lost the case.[62] In April 2014, the decision was affirmed by the U.S. Court of Appeals.[63]

On January 1, 2017, the State of California implemented state bill AB-1687, a SAG-AFTRA-backed anti-ageism statute which requires "commercial online entertainment employment services" to honor requests by their subscribers for their ages and birthdays to be hidden.[64] By the beginning of 2017, IMDb had received more than 2,300 requests from individuals to remove their date of birth from the site. Included in this group were 10 Academy Award winners and another 71 nominated for Oscars, Emmys, or Golden Globes.[65] On February 23, 2017, Judge Vince Girdhari Chhabria issued a stay on the bill pending a further trial, claiming that it possibly violated the First Amendment because it inhibited the public consumption of information. He also questioned the intent of the bill, as it was ostensibly meant to target IMDb.[66]

Birth names publication policy[edit]

IMDb had long maintained that it would keep all valid information, but changed that policy related to birth names on August 12, 2019. IMDb will now remove birth names that are not widely and publicly known, of persons who no longer use their birth names.[67] This was done in response to pressure from LGBTQ groups against the publication of transgender birth names without their consent, which is called "dead-naming". Any name a person had previously been credited under, however, continues to be maintained in the credits section.[67]

See also[edit]


  1. ^"IMDb | History, Features, & Facts". Encyclopædia Britannica. Retrieved October 27, 2020.
  2. ^ ab"Press Room". IMDb. Retrieved August 1, 2021.
  3. ^"Submission guide for users". Help Center. IMDb.
  4. ^Mehta, Ivan (January 11, 2019). "Amazon's new IMDb Freedive service streams ad-supported movies and TV shows in the US". The Next Web. Retrieved January 11, 2019.
  5. ^"Amazon Rebrands IMDb Freedive as IMDb TV, Triples Content". PC Magazine.
  6. ^Dalton, Jeff (March 1, 1990). "Re: Frequently Asked Questions List (2/28/90)". Newsgroup: rec.arts.movies. Usenet: [email protected] Retrieved August 13, 2019.
  7. ^Chmielewski, Dawn C. (January 19, 2013). "Col Needham created IMDb". Los Angeles Times.
  8. ^"Historical Internet Movie Database Site". Cardiff School of Computer Science & Informatics. Archived from the original on March 24, 2013. Retrieved March 21, 2013.
  9. ^"IMDB History". IMDb.
  10. ^"IMDb | Help". IMDb.
  11. ^"Amazon Com Inc – '8-K' for 4/27/98 – EX-99.2". SEC Info. April 27, 1998. Retrieved March 23, 2017.
  12. ^"Internet bookseller announces acquisition of United Kingdom company The Internet Movie Database Ltd" (Press release). IMDb. Archived from the original on June 3, 2012. Retrieved September 18, 2020 – via PR Newswire Europe.
  13. ^Needham, Col (January 1, 2011). "Top 250 contributors for 2010". IMD; announcement. IMDb. Retrieved August 25, 2011.
  14. ^Ng, David (October 19, 2018). "Amazon-owned IMDb ending film festival submission platform Withoutabox". Los Angeles Times. Retrieved June 13, 2019.
  15. ^"Amazon's IMDB Acquires Box Office Mojo; Will Add Box Office Data To Service". CBS News. Retrieved June 13, 2019.
  16. ^"What software are you using to run IMDb?". IMDb. 2017. Retrieved November 16, 2017.
  17. ^Chacksfield, Marc (January 14, 2010). "China blocks number-one movie site IMDb". TechRadar. 2012 Future US, Inc.
  18. ^"The IMDb Studio at Sundance". IMDb. Retrieved December 25, 2020.
  19. ^Sundance 2020(video). IMDb (YouTube).
  20. ^Jones, Craig (April 13, 2017). "The story behind Bristol-based IMDb as they celebrate their 25th anniversary". The Bristol Post. Retrieved May 9, 2021.
  21. ^"IMDbPro signup". IMDb.
  22. ^"Amazon's IMDb launches Freedive, a free movie and TV streaming service — with ads". GeekWire. January 10, 2019. Retrieved June 20, 2019.
  23. ^Hoffman, Harrison (September 15, 2008). "IMDb now serves full-length videos". CNET. Retrieved September 17, 2008.
  24. ^Modine, Austin (September 16, 2008). "IMDb adds full-length streaming movies (Show your US ID card at the door)". The Register. UK. Retrieved September 17, 2008.
  25. ^Spangler, Todd (June 17, 2019). "IMDb Scraps Freedive Name, Will Triple 'IMDb TV' Free Streaming Lineup With Movies From Warner Bros., Sony, MGM". Variety. Retrieved June 20, 2019.
  26. ^Spangler, Todd (February 5, 2020). "Amazon's IMDb TV Inks Disney Pact for Free Streaming of 'Lost,' 'Desperate Housewives,' More". Variety. Retrieved February 18, 2020.
  27. ^Kanter, Jake (July 14, 2020). "'Alex Rider': Amazon Streamer IMDb TV Takes U.S. Rights To Sony Spy Series". Deadline Hollywood. Retrieved August 22, 2020.
  28. ^"Lycos Europe and IMDb sign sales agreement for 9 European markets". Lycos Europe. press release. July 10, 2006. Archived from the original on October 23, 2006.
  29. ^"Can I subscribe only for one month or one year?". Resume FAQ. IMDb. Retrieved January 22, 2008.
  30. ^"Is there any difference between a regular IMDb name page and an IMDb name page created via IMDb Resume?". Resume FAQ. IMDb. Retrieved January 22, 2008.
  31. ^"Copyright and Conditions of Use". IMDb.
  32. ^"The plain text data files". Alternate Interfaces. IMDb.
  33. ^"Which A-list star is hacking IMDb pages?". The Hollywood Reporter. November 14, 2012. Retrieved February 25, 2013.
  34. ^ ab"Java Movie Database (JMDB)". Retrieved October 27, 2010.
  35. ^"Open Media Database". Retrieved December 3, 2015.
  36. ^"Alternate Interfaces". IMDb. Retrieved January 15, 2007.
  37. ^"IMDbPY". IMDbPY. Retrieved February 14, 2011.
  38. ^"Top 250 movies as voted by our users". IMDb. Retrieved June 10, 2015.
  39. ^"The user votes average on film or show X is 9.4, so it should appear in your Top 250 Movies or TV list, yet it doesn't. Why?". IMDb.
  40. ^Norberg, Ragnar (2006). "Credibility Theory"(PDF). Encyclopedia of Actuarial Science. doi:10.1002/9780470012505.tac068. ISBN . Archived from the original(PDF) on May 15, 2004.
  41. ^"IMDB Vote FAQ". IMDb. Retrieved February 6, 2015.
  42. ^"Bottom 100". IMDb. Retrieved March 19, 2020.
  43. ^"Top 250 TV". IMDb. Retrieved March 16, 2016.
  44. ^Each TV episode uses the same message board for the whole series
  45. ^"IMDb is closing its message boards". February 3, 2017.
  46. ^"One of the worst comment sections on the Internet is shutting down".
  47. ^"Should IMDB have preserved its comment boards – flame wars and all?".
  48. ^"Not Even IMDB Is Safe From Trolls". February 21, 2017.
  49. ^"Can someone on the inside explain to me". Customer Community. July 26, 2016. Retrieved March 23, 2017.
  50. ^"How some users are trying to save IMDB's message board". Retrieved September 9, 2017.
  51. ^"The Archiveteam IMDb message board archive (raw data)". Retrieved September 9, 2017.
  52. ^"The MovieChat IMDb Message Board Archive (web-based)". Retrieved September 9, 2017.
  53. ^"17 U.S. Code § 107 – Limitations on exclusive rights: Fair use". Cornell Law School. Retrieved October 8, 2018.
  54. ^"Movie & TV Forums: IMDb Message Board Archives: Are They Legal?". Retrieved October 8, 2018.
  55. ^Bahr, Lindsey (October 18, 2011). "Lawsuit against IMDb revealing private information". Entertainment Weekly. Retrieved April 25, 2013.
  56. ^"Acting unions criticise IMDb in age row". BBC News. BBC. October 29, 2011. Retrieved October 29, 2011.
  57. ^"Actress sued Amazon for revealing age 40 identified as Huong (Junie) Hoang". Sky News. January 7, 2012. Retrieved April 21, 2012.
  58. ^"Actress age claim against IMDb rejected". BBC News. BBC. Retrieved April 12, 2013.
  59. ^Gardner, Eriq (March 27, 2015). "IMDb preserves legal win over revelation of actress' age". The Hollywood Reporter.
  60. ^"Case Docket: United Video Properties Inc., et al v. Amazon.Com Inc. et al.". gov.uscourts.ded.45528. RECAP. Retrieved November 5, 2013.
  61. ^Masnick, Mike (January 12, 2011). "Rovi sues Amazon for not licensing its Electronic TV Guide patent". Techdirt. Retrieved November 5, 2013.
  62. ^Mullin, Joe (November 4, 2013). "Netflix roasts Rovi's 'Interactive TV guide' patents at ITC". Ars Technica. Retrieved November 5, 2013.
  63. ^United Video Properties v. April 8, 2014. Retrieved February 28, 2019.
  64. ^"California enacts law requiring IMDb to remove actor ages on request". The Hollywood Reporter. Retrieved November 15, 2016.
  65. ^Bray, Hiawatha (March 23, 2017). "Internet censorship, Hollywood style". The Boston Globe. Archived from the original on May 16, 2017. Retrieved March 23, 2017.
  66. ^"Judge pauses enforcement of IMDb Age Censorship law". The Hollywood Reporter. Retrieved February 23, 2017.
  67. ^ abMcNary, Dave (August 13, 2019). "IMDb Alters Policy on Publication of Birth Names". Variety. Retrieved March 7, 2020.

External links[edit]

TensorFlow 2.0 Tutorial for Beginners 17 - IMDB Review Classification using RNN - LSTM

How to Build a Neural Network With Keras Using the IMDB Dataset

neural network keras

Keras is one of the most popular deep learning libraries of the day and has made a big contribution to the commoditization of artificial intelligence. It is simple to use and can build powerful neural networks in just a few lines of code.

In this post, we'll walk through how to build a neural network with Keras that predicts the sentiment of user reviews by categorizing them into two categories: positive or negative. This is called sentiment analysis and we will do it with the famous IMDB review dataset. The model we'll build can also be applied to other machine learning problems with just a few changes.

Note that we will not go into the details of Keras or deep learning. This post is intended to provide a blueprint of a Keras neural network and to make you familiar with its implementation.

Table of Contents

  • What Is Keras?
  • What Is Sentiment Analysis?
  • The IMDB Dataset
  • Importing Dependencies and Getting the Data
  • Exploring the Data
  • Data Preparation
  • Building and Training the Model

What is Keras?

Keras is an open source Python library for easily building neural networks. The library is capable of running on top of TensorFlow, Microsoft Cognitive Toolkit, Theano and MXNet. Tensorflow and Theano are the most used numerical platforms in Python when building deep learning algorithms, but they can be quite complex and difficult to use. By comparison, Keras provides an easy and convenient way to build deep learning models.

Keras creator François Chollet developed the library to help people build neural networks as quickly and easily as possible, putting a focus on extensibility, modularity, minimalism and Python support. Keras can be used with GPUs and CPUs and it supports both Python 2 and 3.

Google Keras made a big contribution to the commoditization of deep learning and artificial intelligence since it has commoditized powerful, modern deep learning algorithms that were not only previously inaccessible, but unusable as well.

What is Sentiment Analysis?

Sentiment analysis aims to determine the attitude, or sentiment. For example, a speaker or writer with respect to a document, interaction, or event. It is a natural language processing problem in which text needs to be understood to predict the underlying intent.

The sentiment is mostly categorized into positive, negative and neutral categories. Through sentiment analysis we might want to predict, for example, a customer's opinion and attitude about a product based on a review they wrote. This technique is widely applied to things like reviews, surveys, documents and much more.

The imdb Dataset

The IMDB sentiment classification dataset consists of 50,000 movie reviews from IMDB users that are labeled as either positive (1) or negative (0). The reviews are preprocessed and each one is encoded as a sequence of word indexes in the form of integers. The words within the reviews are indexed by their overall frequency within the dataset. For example, the integer “2” encodes the second most frequent word in the data. The 50,000 reviews are split into 25,000 for training and 25,000 for testing.

The dataset was created by researchers at Stanford University and published in a 2011 paper, where they achieved 88.89% accuracy. It was also used within the “Bag of Words Meets Bags of Popcorn” Kaggle competition in 2011.

Importing Dependencies and getting the Data

We start by importing the required dependencies to preprocess our data and build our model.

(NOTE: You will need to use Numpy, version 1.16.2. Due to a recent change in the framework, Keras has some problems loading the IMDB dataset. Until a new Keras release fixes the issue, this specific version of Numpy will do the trick.)

Continue downloading the IMDB dataset, which is, fortunately, already built into Keras. Since we want to avoid a 50/50 train test split, we will immediately merge the data into data and targets after downloading so we can do an 80/20 split later on.

Exploring the Data

Now we can start exploring the dataset:

You can see in the output above that the dataset is labeled into two categories, — 0 or 1, which represents the sentiment of the review. The whole dataset contains 9,998 unique words and the average review length is 234 words, with a standard deviation of 173 words.

Let's look at a single training example:

Above you can see the first review of the dataset, which is labeled as positive (1). The code below retrieves the dictionary mapping word indices back into the original words so that we can read them. It replaces every unknown word with a “#”. It does this by using the get_word_index() function.


Sign up for free to get more Data Science stories like this.

Data Preparation

Now it's time to prepare our data. We will vectorize every review and fill it with zeros so it contains exactly 10,000 numbers. That means we fill every review that is shorter than 10,000 with zeros. We need to do this because the biggest review is nearly that long and every input for our neural network needs to have the same size. We will also transform the targets into floats.

Now we split our data into a training and a testing set. The training set will contain 40,000 reviews and the testing set 10,000.

Building and Training the Model

Now we're ready to build our simple neural network. We'll start by defining the type of model we want to build. There are two types of models available in Keras: the sequential model and the model class used with functional API.

Next we simply add the input-, hidden- and output-layers. Between them, we are using dropout to prevent overfitting. Please note you should always use a dropout rate between 20% and 50%.

We use “dense” at every layer, which means the units are fully connected. Within the hidden-layers we use the relu function because this is always a good start and yields a satisfactory result most of the time. Feel free to experiment with other activation functions.

At the output-layer we use the sigmoid function, which maps the values between 0 and 1. Note that we set the input-shape to 10,000 at the input-layer because our reviews are 10,000 integers long. The input-layer takes 10,000 as input and outputs it with a shape of 50.

Lastly, we let Keras print a summary of the model we have just built.

Now we compile our model, which is nothing but configuring the model for training. We use the “adam” optimizer, an algorithm that changes the weights and biases during training. We also choose binary-crossentropy as loss (because we deal with binary classification) and accuracy as our evaluation metric.

Now we're able to train our model. We'll do this with a batch_size of 500 and only for two epochs because I recognized that the model overfits if we train it longer.

The batch size defines the number of samples that will be propagated through the network and an epoch is an iteration over the entire training data. In general, a larger batch size results in faster training, but doesn't always converge as fast. A smaller batch size is slower in training but it can converge faster. This is definitely problem dependent and you'll need to try out a few different values. If you start with a problem for the first time, I recommend first using a batch-size of 32, which is the standard size.

It is time to evaluate our model:

Awesome! With this simple model, we already beat the accuracy of the 2011 paper that I mentioned in the beginning. Feel free to experiment with the hyperparameters and the number of layers.

You can see the code for the whole model below:

You can now use this model to also do binary sentiment analysis on other sources of text but you need to change them all to a length of 10,000 or you change the input-shape of the input layer. You can also apply this model to other related machine learning problems with only a few changes.

Niklas Donges is an entrepreneur, technical writer and AI expert. He worked on an AI team of SAP for 1.5 years, after which he founded Markov Solutions. The Berlin-based company specializes in artificial intelligence, machine learning and deep learning, offering customized AI-powered software solutions and consulting programs to various companies.

RelatedRead More About Data Science


Imdb network

The government pretended to tolerate his antics, because he did not rob the city and honestly pay the bills, but in fact it was Matthew. Who tolerated the power, because his people needed somewhere to have fun. Of course, Matthew knew from the fact that newcomers had come to town.

Practical implementation of Neural Network for Text data - IMDB review data - Ronak Patel

And who said it was a sin. And who are the judges. Neighbor Varka is an alcoholic from the third floor. How can love between mother and son be a sin. We do not sleep, after all.

You will also be interested:

To my uncle Andrey, his daughter Katya and wife Olya. They greeted me very hospitably, settled in my room, which was next to Katina. The room was spacious, with its own bathroom and a large bed. The only thing missing was a laptop, without it a week would seem like an eternity.

1911 1912 1913 1914 1915