Words: 2712
The software as a service platform is used for tagging, describing, categorizing, annotating, labeling, and organizing online media and other forms of knowledge work. During my time as an monitoring, task creations, Mature, Pinsetters, Google Refine, and Search Relevance. So what is Search engine optimization? It is defined as a process of improving search relevance off website from web users. Search engines have been established in the worldwide network since the birth of the Internet. But many users had problems with these search engines, so webmasters began researching on how to better optimize hem.
Since the mid-sass’s until now, new companies such as Thesauri are researching how to implement an automatic software that improves the relevance of what a user wants to search. For over 15 years, these companies have found methods to help clients by learning how search engines rank pages. SEE may target different kinds of search that organize a wide variety of media content such as images, local search, videos, blobs, and etc. Thesauri’ technology combines algorithms with human and collective intelligence to solve complex problems that cannot be solved with either component alone.
The company help organizations scale their business process by decomposing knowledge-intensive tasks into workflow comprised of discrete micro-tasks that are powered by humans and machines. This provides increased productivity, reduced costs and better quality leading to new opportunities for engagement. Thesauri’ platform manages the full stack: content, workflow, workers, pricing, payments, quality and integration so you don’t have to. Thesauri has been featured in The New York Times, Wired, Business Week, The Economist and a host of others. The National Endowment for the
Humanities awarded Thesauri and The Museum of the City of New York a grant to annotate the museum’s archive. When I initially started with Thesauri, I spent countless hours with business managers to understand the processes of how this product works. This Job has taught me the knowledge of starting and developing a “real world” software. Throughout the course of my college education, I have learn various concepts of how business functions. More specifically, the logistics of how a company flows. I also have some computational knowledge with C++ and have utilized these concepts in some of my assignments with Thesauri.
To have a better grasp of how the company functions, I had to understand the product that they were selling. “Tasks”: Companies and individuals use Thesauri to help improve search engines for their websites. Thesauri uses research to find efficient ways to create tasks that will exploit the most optimal way to tag media content. By creating tasks that are powered by humans and machines, we could provide quality tags. So what is an example of a task? Well one of the task that Thesauri recommends to efficiently tag an image is discovering what the photographic styles are in the image.
This task squires a human to select tags that are relevant to the image. Some of the tags are implemented ask whether the photo is taken from an aerial view or a low-angle view or if it’s a scenic photo, etc. One of the main reasons behind creating task is that Thesauri wanted to combine both automation and human to optimize metadata tagging. One of my tasks was to actually go on Mechanical Turk and help tag these images individually. I would spend countless hours of doing these tasks so that I can understand the processes of how the product works.
I believe that this experience has taught me to really think like a yester engineer. When doing this task, I used my knowledge of flow-charts and quickly analyze the inn’s and outs of what Thesauri is all about. Overtime, I acquired the technical skills to analyze the logic behind the algorithm of these tasks. I was able to use my prior knowledge of systems and apply it to test the efficiency of this task. To elaborate more about the product, as mentioned earlier, it requires humans to choose the tags for these “task”. So how do we get workers to do these tasks?
Thesauri specifically uploads this task onto Amazon’s Mechanical Turk where rocker (humans) can choose the right tags under the selected images. Mature is an online marketplace for temporary workers to work on “task ” that computers can’t do efficiently. To create an incentive for these workers to do your task is that a worker can receive up to a minimum of 5 cents. Workers-people from around the world looking to make a little extra money from home can spend their spare time doing these tasks. So I would physically make an account on Amazon and go on Mechanical Turk to work on these tasks.
While working on these tasks, I would also try to find flaws with these ask creation. Flaws such as if an image did not load on the Thesauri platform correctly, if any alignments are off with the task, etc. I was responsible to report any flaws that I saw to our company’s debuggers via email. I was also in charge of the back end of the website, monitoring the quality of tags that these workers did for the tasks. More specifically, I was in charge of the Quality Assurance of the task creations. I was given an admit account to access and monitor workers that worked on our tasks on mechanical Turk.
If a worker was producing bad-quality tags for the task, I old physically message the worker and tell them what they were doing wrong. Sometimes it could get really confusing when doing these task, so by having someone giving them feedback will optimize the quality of these tags. This made me feel important to the company. By giving me the responsibility to actually monitor these workers made me realize how crucial it was to do my Job correctly. By doing this, it helped me get experience with dealing with “real-world” applications such as giving feedback to people’s work.
Research: Another area that I was apart of is the research of Search Engine Optimization. I specifically used the website Pinsetters to research how users were searching the that they like onto their account, which is similar to “liking” a photo on Backbone. So our product manager was interested in why users were repining certain images. Was it because of the tags? Or was it because of something else? So I experiment with uploading certain pictures onto Pinsetters and applied some of the Thesauri tags to these images.
We applied thesauri tags because we hypothesize that users are more likely to repine a picture if they can easily search them on the engine. What I mound was that users did not repine picture because it was the pictures could be easily search. They simply Just repine them because it was visually appealing to them. So our results did not align with our hypothesis with users liking an image because it could easily be searched. What I took from this experience was learning how to think outside the box. As an engineer, we are going to face many problems in our lives that require us to think outside of the box.
So by utilizing my ability to think outside the box, I can effectively figure out the reasoning behind how things work. Although my mime with Thesauri was an enriching experience, I was not able to get involve with the complexities of the technicalities of these tasks. I did not get to utilize my technical abilities to its fullest potential. For instance, I did not get to experience any computer coding with developing the algorithm to our system. Since I did not have adequate computer science knowledge, I was left out of being apart of developing the code to our software.
However, I still felt apart of the software-side by constantly monitoring any mistakes I found with the algorithm. An example of a typical mistake I old find is a common coding mistake with a text. In C++, Nan’s or it’s ” are codes to that allow the text to create a new line. If you look at the picture below, you can see that this code did not work on the software correctly. I would report any mistakes like this to our computer programmers via email. Search Relevance: But on the contrary, there was one project that required me to utilize my knowledge from my education to finish an assignment.
I was asked to use my knowledge of statistics and probability to this project. The project that I was in charge of was the search relevance project. One of Thesauri’ main customers, Magnum Photos, was not completely satisfied of the tags that were produced for the images. The problem was that the tags that Thesauri provided for them were not popping up on the websites when the user searches the term. This was a major problem because it pointed out the flaws of the quality of the tags. The relevance of a tag was not as optimal as it should have been.
So my Job was to statistically analyze the relevance of these tags. From the task, there are certain tags that are considered to be “Thesauri tags”. So what I was responsible for was to go on the Magnum Photo website and heck out the relevance of each tag. I would physically type the tag into Magnum Photo’s search engine and see if images correctly correspond to the tags. I would then categorize the results into three categories: relevant, not relevant, and maybe relevant on to an Excel spreadsheet. Then I computed the percentage of how many picture were relevant from the first one hundred photos.
To give examples consider the term “Day’. I typed the word into the Magnum Photo’s search engine and thousands of photos would pop up. I looked at the first one hundred photos and identify if the images were pictures of “day’ or not. So for this term I found that 76/100 of the photos were relevant, 8/100 of the photos were not relevant, and 16/100 are maybe relevant. I would repeat this process for the rest of the tags that I had to check. The reasoning behind this project is to find which tags would create the most ambiguity. Sometimes it is hard for a worker to identify some of the ambiguous tags.
For example tags that determine what age group are in the image. A problem with this was to identify a young adult compared to a middle-aged adult. Although there are some key differences, it is hard to actually objectify whether if the person in the stricture is a young adult or a middle-aged adult. From doing this search relevance project, I was able to utilize my engineering skills to analyze and compute the data. The data told us which terms were not producing the most optimal results. One of the areas of tagging that had a low percentage of relevance was the photo orientation tag: Horizontal, Square, Vertical, and Panoramic.
What I found out that photos orientation tags could only produce 46 percent relevance. Results like this were extremely helpful for Thesauri to understand. By quantifying the search relevance, Thesauri was able to present these founding to Magnum Photos. In order to implement quality assurance, we need to figure out which areas of our tagging are not optimal. This assignment really taught me the importance of statistical analysis. By using the knowledge that I learned for my stats/probability class, I was able to understand which areas of tagging are causing problems.
Conclusion: My first semester as a co-pop intern to Thesauri was an easy experience because my managers and others helped guide me through the process of understanding the projects that I was in charged of. My daily communication with the Thesauri team has made me realize the importance of communication. Without it, teams can really be dysfunctional and things will not get done in an efficient manner. Communicating with the team was extremely important to get us updated with current SEE findings that people are researching and informing me of any important deadlines.
When I had any problems or questions, I was always able to get a quick response or answer that would resolve it. Our main method of communication was through Spiky. So I would go on Spiky to see if any of the business managers were online to answer my questions. Since I was actively involved in contacting my managers, I was able to complete my assignment at an orderly fashion. I also worked with another internee, so communication between both of us was very crucial. If he could not complete a certain deadline because he sick, he had to message me to let me know that he did not finish his part.
This was crucial because there was an instance when this happen. A customer was expecting us to finish a certain assignment and the other internee was under the weather to complete it. So he was able to reach me quickly on Spiky and told me that he couldn’t work on the assignment and asked me if I could work on see, communication is extremely important to getting things done under time constraints. From this experience, I now understand the importance of communication. Finally, I had a phenomenal experience with my co-pop internship with Thesauri because I was apart of the start of the company.
I felt like I played a crucial role in the developing stage of the product. By ensuring quality from my monitoring responsibilities, I was able to analyze the processes in order to maintain an efficient product for customers. I learned the stages of starting a big-company like Thesauri. By working closely with the business managers such as the CEO and the COO, I was really exposed to how a business functions. I now understand what it sakes to effectively get things done in an orderly timed fashion. The biggest thing I learned from the business-side is how a CEO effectively communicates to his company.
With a company like Thesauri, every department needs to be on the same page. So I got to see how our CEO communicates to the software developers, product managers, marketing team, research team, etc. Sometimes it is hard for the CEO to motivate his company because the starting stage is long and painful. But with Thesauri, I was able to see how our CEO was able to use influence tactics to persuade our team into getting things done. So I am very appreciative of getting ands on experience of seeing how a company such as Thesauri functions.
Thesauri has given me the tools to understand how to start a company. Hopefully, one day I can use this knowledge to start a company of my own. Overall, I am glad that I took this internship opportunity with Thesauri. With my months with them, I have learned so much about what the real world is like and what it takes to run a company. A hand on experience like this can really prepare me to face the professional world. I am very thankful to participate in the co-pop experience at USC. By working with my business managers, I have built a good chemistry and legislation with them.
This could hopefully benefit me in my future endeavors with my career in engineering. After this experience, I feel like I am very qualified and equipped to Join any company I plan to pursue. By using my knowledge that I learn from Thesauri such as my engineering skills, I will have a better eye to effectively analyze any problem. My plan for the future with Thesauri is to perhaps still work with them on side of my other Job. Since I have built good relationships, I will stay in touch with them and hopefully one-day cause a bigger impact on the company.