AI-Based Solution Developed for Measuring Emotional Climate of NY

Domantas Didziapetris, a young scientist from Kaunas University of Technology (KTU), Lithuania has created an artificial intelligence based solution for computing the emotional climate in Manhattan Island, New York. After conducting a sentiment analysis of more than 36 thousand tweets, he developed a scale which shows how the residents and visitors rate emotional climate of the various Manhattan neighborhoods.

(Image credit: Kaunas University of Technology)

One of the world’s most-visited cities is New York. According to the official statistics, in 2018, New York City received a record 65.2 million visitors, and the numbers are persistently increasing. Manhattan, the most thickly populated of the five boroughs of New York City, with a population of 2.7 million and draws nearly 15 million tourists every year.

“Manhattan is the heart of New York City, bursting with life. And what could be a better place to test a new methodology if not the city which literally never sleeps?”, says Domantas Didziapetris from KTU Faculty of Architecture and Civil Engineering.

According to the young scientist, urban analysis approaches frequently still use paper-based or online surveys. Yet they are not constantly dependable, and the results may not signify the real situation.

One glance at a social media platform may provide more insight into the actual situation than an in-depth questionnaire. Moreover, the opinions expressed here are far more expressive and impulsive, as people tend to express joy or disappointment online the very moment they feel it. One of the best platforms to gather such data is Twitter as it limits the number of characters in a message, it is easy to depersonalize the data and the agreement to share the data is already granted once a user signs up to Twitter.

Domantas Didziapetris, Architecture and Civil Engineering, KTU Faculty

Some IT programs were written so as to process the data. Firstly, Python was used to assemble software that would collect the tweets in real time. During a four-month period, over 1 million tweets were collected, 65.447 of them had geographical coordinates. After removing all the entries that were from outside of Manhattan, 36.543 tweets were examined in detail, categorized by the community district boundaries as identified in the New York City Open Data portal.

Secondly, the scientist developed a program for sentiment analysis of the tweets. Two criteria were taken into consideration – subjectivity and polarity. The subjectivity points to the factual content of the tweet and the polarity – its emotional nature. Both of the criteria can be graded from -1 to +1.

“The higher the subjectivity criteria, the less reliable is the entry. The closer the polarity is to -1, the less positive is the emotional tone of the tweet”, explains Didziapetris.

The sentiment analysis results were laid out on the map, and, according to the young scientist, this plainly exposed which neighborhoods are, as it were, happy, and which ones – were not. Upper West Side had the least positive emotional climate in Manhattan.

Upper West Side is still very big an area to derive conclusions; so, it was essential to pinpoint the actual location which would require reviewing and urban development.

To finish this task, the “Urban Network Analysis Toolbox for ArcGIS”, a toolkit developed by MIT in 2011, for urban network analysis was employed. One of its techniques – betweenness – is usually used to calculate and to estimate the potential of passersby in the network. When the area is more reachable, then the color on the map is warmer (red).

“The less reachable the area, the more abandoned it is, and the abandoned areas are usually more prone to criminal activity. After urban network analysis, one of the areas in the neighbourhood looked especially suspicious – after the visualization, it turned blue and green. My hypothesis was proven right after I received the photos from the area. Simply put, it lacks safety. My supervisor, who took the photos, described the common feeling in the area as ‘unsafe’, ‘not cosy’, ‘encouraging to leave as soon as possible’, ‘old’. The really interesting thing is that this territory is nearby active and reachable buildings in the East, such as Lincoln Center, The Juilliard School”, says Didziapetris.

The study, which was presented in Domantas Didziapetris Master’s Project, was awarded by Hnit-Baltic company, Esri Inc. representative in the Baltic States. Domantas Didziapetris has won a chance to showcase his research to an international audience of 18 thousand people at the Esri User Conference 2019, which will happen in San Diego, California in July.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this news story?

Leave your feedback