Part 1: Our Sweet Home
Part 2: Love In 12 Minutes
Part 3: Machiavellian Berkan's Concerns
Part 4: Bonanza and His Friends
Part 5: Long-lasted Love
Part 6: Pre-design Battle
Part 7: My Dear Father
Part 8: Video Speech of A Public Enemy
Part 9: A Job Interview
Part 10: The Second Great Attack
Part 11: Big Hopes
Part 12: An Unexpected Disaster
Part 13: Sediment of the Days
Part 14: The Goddess Of Matchmaking, Gülizar
We could not reach a full agreement on the name Gülizar: “At first glance, it seems like a good idea, but something's missing,” said Deniz.
Berkan criticized the name more brutally: “I don't know if you are ally or enemy, brother. What is Gülizar, we get the stamp of outdated as soon as people hear it.”
We were working at my house because we didn't have the money to rent an office. I got up and started drawing the cartoon of Gülizar on the writing board at the corner of the Hall. I made up this name inspired by the aunt Gülizar, who runs the residence of Berkan, so I drew a woman who tied her head scarves under her chin and added a lens to the left eye of the woman in the caricature and a microphone to her face from her left ear to her mouth.
”When processing commands given by Gülizar users, spiral light reflections will take place in the lens on the left eye," I said.
“You drew it bad, but now I understand the idea better,” said Deniz.
”I am not satisfied by the name, but maybe if it is presented like this, " said Berkan, gradually warming up to the name of Gülizar.
To relieve Berkan, I said, “the name is up to a point, the function of the platform that matters.”
“Is there anyone who can draw this animation? At the stage where we will present a suitable partner to users, Gulizar's head can be completely turned,” said Deniz.
“We will find one, probably drawing a single cartoon would be enough, the lens in the eye is not very difficult to revive, it is easy to turn the head around on its own,” I said. After roughly creating the character to represent our platform, it was time to create performance metrics. “How will we measure the success of our platform?” I asked.
“If people come often, we're successful,” said Berkan.
“Why should people come? They have no other occupation?” I asked him.
”Don't ask stupid questions, don't spoil my temper," said Berkan, who was probably upset because I was not satisfied with his answer.
“A longlasting relation is achieved through five stages; awareness, meeting, recognizing, liking and being together. We hope that the love algorithm will help us to make people aware of the people that fit them. Once they have passed this stage, they will send a message to each other on the platform to meet. To get to know each other closely, we will open the detailed profiles of the couples before the virtual interview. Couples will get to know each other in virtual interviews by taking advantage of the detailed profile information. At this stage, they may decide not to meet again, continue to have a virtual interview or choose to meet at an outside location. We can assume they like each other in the scenario where they continue to interact. As a result of this mutual appreciation situation reaching a certain intensity, unity will be formed. The number of users reaching each of these stages is important to us, I think,” said Deniz.
Berkan turned to Deniz and asked, “Do you have a boyfriend?”
"It is none of your business,” Deniz answered.
“You don't want to say it, so you don't have somebody to mention. If you had a boyfriend, we'd have known by now.”
“How can you be so sure, maybe he's far away,” said Deniz.
“If so, why are you getting angry?” asked Berkan.
“I'm not angry; I just don't want to talk about it.”
“Two of the three partners don't have lovers, and the third one doesn't even want to talk about it, and these three young people will solve the problem of love in the country,” I said.
"If the bald had ointment, he would spread to his head," said Berkan.
“The need is the mother of discovery; we'll find our lovers on our platform,” I answered.
”I wonder if we are trying something we can't accomplish," said Berkan.
“People who set up successful platforms are like you and me. We don't need to rush. Be brave,” said Deniz.
Berkan said in a sarcastic tone, “Orders understood my sister, we will attack the enemy with determination.”
”That's precisely what you're going to do, " said Deniz.
After this little debate, each of us turned around and started dealing with things that were in our fields of expertise. I continued to develop the love algorithm, which would be the core of the platform. Artificial neural networks have been used everywhere that requires “intelligence” in the past decade, making a crazy leap. These algorithms, similar to neural networks in our brains, find and extract the patterns from the data so that it was possible to solve the complex problems without the need to write a unique identifier program. Thanks to various versions of these algorithms, they can be converted to speech and writing, handwriting can be read, perfect translation from any language can be made, and objects within pictures and videos can be recognized. The process of choosing people's ideal partners contained quite complex dynamics. During the procedure, information feedbacks disrupted the harmony of equations, creating significant computational difficulties. The secret was to have a vast database and enough processor power. A few days ago, I had access to Facebook data of 300,000 people created for researchers. When I downloaded the data to our servers, I felt the enthusiasm and excitement of a pirate who had found a treasure, and I worked all night to satisfy my curiosity. When I found out that the first models I built had predicted the couples that were in contact with each other by more than 90%, my heart was pounding as if it were going to bounce off. I never expected such easy success, and since it was so easy, I knew there was a mistake. I left out variables such as common Last Name, shared children, common friends, etc. that could be used to predict the couples. I went to bed because I had difficulty interpreting the results that were still too good to be true.
Five hours later, when I woke up in full sleep and went to the computer, I found the answer. Artificial neural network algorithm detected pairs because they were on the same images or videos. After a couple started to be together, they left dozens of traces of being together in the social media, and it was not useful to make a successful guess on these tracks, and for healthy modeling, there was a need for data that showed their status before the time they started to be together. Thus, the model would have to focus on real variables that would support the prediction of relationship. I realized that if I focus on couples who have started to be together in a certain period, and I guess through their situations at the beginning of that time, I might discover the real love algorithm. In fact, my intention was not original, but everyone who works in the field of data analytics knows the risks of modeling through a single time segment. With the excitement of having achieved a full set of data, I forgot to ask for the previous social media data of the same people. As soon as I realized the situation, I applied to the institution and received a positive response. But I had to wait another ten days to get comparative data.