There are 220 million speakers of Dravidian languages in the world. Most of these are speakers of Tamil or Telugu who live in South and South-Central India. That population is comparable to the English speaking population of the world, of which 360 million are native speakers. As such it seems reasonable to expect that speakers of Telugu and Tamil should have a notable presence in the world of cinema. This part of cinema is sometimes known as Tollywood.

In fact, from our previous analysis of the IMDb professional graph, we have already seen some of the power of Tollywood. The number person with second highest PageRank is Brahmanandam, an actor, the majority of whose titles have been Telugu language productions. In that previous post, I made the argument that the greatest actor of all times was Matsunosuke Onoe, the prolific first superstar of Japanese silent cinema who had the highest PageRank. Please, now, let me make an argument for why Brahmanandam is the greatest actor in the history of all times.

Consider again our dedicated cinema buff who helped demonstrate the PageRank algorithm. In the previous scenario, she was willing to browse through the professional graph forever. At each moment, there was a small chance that she would become bored with her current browsing and choose some node uniformly randomly from all nodes in the graph. Otherwise, she would choose uniformly randomly a node connected to the current node she was browsing. If she was looking at a person, she would look next at a movie from that person’s filmography. If she was looking at a movie, she would look next at an actor, writer or director who contributed to that movie.

The PageRank of a node in our professional graph is the probability that after browsing the professional graph for an infinite amount of time, the dedicated cinema buff will be looking at that node. The procedure so far is easy enough for the cinema buff, but it does not allow her any personal interests or preferences. There is no particular reason why she would choose to look at one movie or person as opposed to another.

Does it not seem reasonable that we should allow the dedicated cinema buff the latitude to follow connections which seem more interesting to her than others. Perhaps we might allow her to prefer movies with higher ratings or more votes. Perhaps we might allow her to prefer actors that appear first in the movie credits. All of the information she would need to make these preferences are available in the graph we have defined from the IMDb data. Hence, let us mark the edges of the graph with weights defining these possible preferences, so that she can use those weights to follow some more interesting connections with higher probability than others.

The PageRank algorithm works just the same way on these weighted graphs as it does on the unweighted graph from earlier. For each of these weighted versions of the graph, the PageRank of a node in the graph will be the probability that after an infinite amount of time, the dedicated will be looking at that particular node. For each of these weighted graphs, these PageRanks will be different, and so some movies or people will be ranked higher than they were in case of the unweighted graph.

There is some arbitrariness in how to choose these weights based on interesting features. For the experiment I describe here, I chose functions which mapped features to numbers between zero and one. Ratings were easiest, because those are numbers between zero and ten, so it was only necessary to divide by ten to get a weight. For the others, I chose mappings on a somewhat ad hoc basis so that the weights had good appearance when plotted in a histogram.

For the weights thus defined, here are the top 10 people according to PageRank.

UnweightedRating and votesCredit order
0Matsunosuke OnoeBrahmanandamMatsunosuke Onoe
2Kinya OgawaMohan BabuAdoor Bhasi
3Sakae NittaWilliam ShakespeareSung-il Shin
4Jirô YoshinoMammoottyMohanlal
5Floyd ElliottMohanlalMammootty
6Michael BlackwoodRajkumarShivaji Ganesan
7Minoru InaoCüneyt ArkinFloyd Elliott
8Satoru KobayashiRon JeremyCüneyt Arkin
9PandharibaiFrederick WisemanBahadur
Top 10 by PageRank of whole graph with different weights

The first column shows people by PageRank for the unweighted graph, the second column for the graph weighted by ratings and votes of movies, the third column for the graph weighted by cast order in credits. While all three columns show high ranks for obviously famous people, the second column is the one which rates William Shakespeare the highest. There are few people in all of history who are more famous than William Shakespeare, and so any good measure of fame should rate him highly.

While this measure of fame throughout the entire world of cinema must be meaningful in some sense, there is no one in the world who could be equally familiar with all markets everywhere. It is not humanly possible to watch every movie that has ever been made and understand them all in their many languages. Hence, to get a better understanding of what these different ranks mean, consider a more limited population of movies and movie makers, such as those within a professional distance of a particular actor. Consider the ego graph centered around a particular person, that is the graph whose vertices are at most a given distance from that person. In the world of English speaking cinema, the traditional ego graph is that around Kevin Bacon. Let us look first at those people who have participated in creating some movie with Kevin Bacon. These are the people with professional distance at most one from Kevin Bacon (those with Bacon number 1 or less).

UnweightedRating and votesCredit order
0Brian GrazerRobert De NiroRobert De Niro
1Jason BlumJason BlumJeff Bridges
2John DavisClint EastwoodTom Hanks
3Tim BevanJeff BridgesClint Eastwood
4Eric FellnerTim BevanMeryl Streep
5Brad PittJohn MalkovichAlec Baldwin
6Robert De NiroMeryl StreepBrad Pitt
7Ron HowardJoe MantegnaMichael Douglas
8Tom HanksAlec BaldwinRyan Reynolds
9Jeff BridgesBrian GrazerTom Cruise
Top 10 by PageRank of those with Bacon number at most 1

To me, many of these names are familiar in all three columns. When the graph is weighted by either credit order or ratings and votes, Robert De Niro is top ranked. I do feel that he is very famous, so both rankings are candidates to measure fame.

According to rank for the graph weighted by rating and votes, Clint Eastwood is ranked third. According to rank for the graph weighted by credit order, Tom Hanks is ranked third. Whether one of these is more famous than the other is something that their various fans might dispute vigorously, but, again, both are clearly famous people.

Let us look a little bit further to those people with Bacon numbers at most 2.

UnweightedRating and votesCredit order
0William ShakespeareWilliam ShakespeareEric Roberts
1Bülent OranPaulo BrancoJoseph Estrada
2Eric RobertsBülent OranJohn Wayne
3Paulo BrancoEric RobertsKen Takakura
4Eckhart SchmidtArtur BraunerAkshay Kumar
5Prem ChopraAndy WarholGérard Depardieu
6NassarAnupam KherShashi Kapoor
7Anupam KherPrem ChopraMichael Madsen
8Gulshan GroverMenahem GolanFranco Nero
9Andy WarholNassarFrancesca Bertini
Top 10 by PageRank of those with Bacon number at most 2

That small distance from Kevin Bacon is already enough that this population has started to look pretty multi-cultural. Note, however, that for rating and votes, already, William Shakespeare has appeared. Again, I feel pretty confident that any measure of fame should rank Shakespeare highly.

Let us look at the ranks of people with Bacon numbers at most 3.

UnweightedRating and votesCredit order
1Michael BlackwoodSrideviSung-il Shin
2PandharibaiMohan BabuMohanlal
3William ShakespeareWilliam ShakespeareMammootty
4Wui NgMammoottyShivaji Ganesan
5Eddie GarciaMohanlalCüneyt Arkin
6Sung-il ShinRajkumarBahadur
7SrideviCüneyt ArkinMichael Blackwood
8Krishna GhattamaneniRon JeremyMithun Chakraborty
9MammoottyFrederick WisemanBalkrishna
Top 10 by PageRank of those with Bacon number at most 3

Ah. Now is that not interesting? Here, at this distance from Kevin Bacon, our hero of cinema, Brahmanandam ranks top by all three measures.

Those of you already familiar with Telugu cinema need no further convincing, I am sure. For those to whom this is new, Brahmanandam is a comedian who currently holds the Guiness Book of World’s Records place for most screen credits for a living actor, having appeared in over 1000 films. He is also one of the top paid actors in Telugu cinema, often charging by the scene. His career spans decades and continues to this day.

To get a better idea of his skills, please watch a few of his movies. Consider watching, for example, Prayanam, a romantic comedy taking place almost entirely in an airport, or Eluka Majaka in which Brahmanandam plays a mouse god who is offended by the romantic hero. Still, if after watching those, you are not convinced, then you must witness Brahmanandam’s portrayal of “Kill Bill Pandy” in Race Gurram. Seeing that movie will remove any lingering doubts you may have about my reasoning that Brahmandam is indeed the greatest actor in all of modern cinema.

The code to compute PageRanks for IMDb data set