Author: SocialXYZ

I noticed something interesting about Waze the other day, which confirmed some suspicions I had for a few weeks.

For some time, visitors keep getting lost getting to my house, which sits on a cul de sac near the Los Angeles coastline. On more than one occasion, people got lost when using the Waze app, which is supposedly the best crowdsourced navigation around. The app would direct them down deadend streets, blocks away from my home. As it turns out, this may be my fault (not really, but really)…

I witnessed this phenomenon first hand the other night returning from a wedding at the Aquarium of the Pacific. It was about midnight that I noticed my Uber driver turning down a deadend street near my home. When I asked where the driver was going, the directions stated, in 450 feet, make a left on unnamed street. He proceeded to miss the turn.

This is the route I usually take, but I was surprised that Waze would recommend this to an Uber driver. The road is unnamed because it’s not a road, but the back alley of my townhome’s complex. The back alley connects two streets by about a 1/4 mile stretch marked with speed bumps and blindspots.

My suspicion is that because I take this route so frequently and Waze tracks me so closely, Waze thinks that this is now the optimal route. Problem is that it’s only optimal if you know about the alley, not if you miss the turn. Go figure…

Read More

Here is the Wordle.net word cloud for work I co-authored, titled, “Affordances of Recommender Systems for Disorientation in Large Online Conversations,” which was published in the Journal of Computer Information Systems. This work is another collaborative effort with Asst. Professor Evren Erylimaz.

ABSTRACT
In the context of large annotation-based literature discussions, this research examines the affordances of recommender systems on users’ disorientation. Drawing insights from literature on group cognition, knowledge building, and recommender systems, we developed three recommender systems and tested these systems on 136 users. Results indicate that the recommender system with constrained Pearson correlation coefficient similarity metric reduced users’ disorientation and afforded them the opportunity to become better aware of interesting and relevant information based on their needs and preferences without heavy costs in terms of time and effort. With respect to other software conditions, results indicate that users suffered from higher levels of disorientation. These findings counter the claim that annotations reduce disorientation. Theoretical and practical implications are also discussed.

E. Eryilmaz, B. Thoms, Z. Ahmed, KH Lee, “Affordances of Recommender Systems for Disorientation in Large Online Conversations,” Journal of Computer Information Systems, April, 2019. (Impact Factor 1.56)

Read More

Here is the Wordle.net word cloud for work I co-authored, titled, “Development of a Reading Material Recommender System Based On Design Science Research Approach,” which will be presented at the 52nd Hawaii International Conference on System Sciences on January 10, 2019. This work is a collaborative effort with Asst. Professor Evren Erylimaz.

The paper abstract is as follows:

Using design science research (DSR), we outline the construction and evaluation of a recommender system incorporated into an existing computer-supported collaborative learning environment. Drawing from Clark’s communication theory and a user-centered design methodology, the proposed design aims to prevent users from having to develop their own conversational overload coping strategies detrimental to learning within large discussions. Two experiments were carried out to investigate the merits of three collaborative filtering recommender systems. Findings from the first experiment show that the constrained Pearson Correlation Coefficient (PCC) similarity metric produced the most accurate recommendations. Consistently, users reported that constrained PCC based recommendations served the best to their needs, which prompted users to read more posts. Results from the second experiment strikingly suggest that constrained PCC based recommendations simplified users’ navigation in large discussions by acting as implicit indicators of common ground, freeing users from having to develop their own coping strategies.

And below is us at the conference site. Not a bad view at all.

Citation
E. Eryilmaz, B. Thoms, KH Lee, M. de Castro, “Development of a Reading Material Recommender System Based On Design Science Research Approach,” Proceedings of Hawaiian International Conference on System Sciences (HICSS 52), January 8-11, 2019, Wailea, HI, USA.

Read More

This Saturday I’ll be flying off to present the co-authored paper, Dynamic Visualization of Quality in Online Conversations in Santiago Chile at the 2018 IEEE/WIC/ACM International Conference on Web Intelligence. I’m very excited to present this research because it was a truly collaborative effort with 3 undergraduate students and my long-time collaborative Dr. Evren Eryilmaz. Below is the abstract and tagcloud of the paper.

Abstract— This paper reports on software designed to visualize levels of quality within online conversational media. Prior to construction, data mining was performed on 2,157 online conversations and examined for attributes of quality. This initial dataset was analyzed for lexical complexity and prompt-specific vocabulary usage and helped guide the redesign of an existing asynchronous online discussion board (AOD). The new design incorporates a real-time quality analyzer and provides users with a visual breakdown of their post in relation to the overall group discussion thread. Results found that the proposed system produced higher levels of overall quality in discussion posts and increased interactions with higher quality discussion posts. Survey results and a social network analysis (SNA) indicate that the proposed system produced higher levels of system satisfaction and group cohesion when compared against control software.

Citation
B. Thoms, E. Eryilmaz, N. Dubin, R. Hernandez, S. Colon-Cerezo. “Dynamic Visualization of Quality in Online Conversations,” Accepted for inclusion in Proceedings of 2018 IEEE/WIC/ACM International Conference on Web Intelligence, December 3-6, 2018, Santiago, Chile.

Read More

I recently moved from my nice cozy 1-bedroom apartment, to a 3-bedroom, 3-bathroom townhouse. With the transition came a wave of new technological additions including a Google Home Max for the living room, a Google Home for the upstairs, a Google Nest camera for the living room, and Google Nest thermostats for upstairs and downstairs climate zones… Oh, and I also picked up four smart lights (lifx) and adapters. Now for the quick good, bad and scary of moving to IoT world.

The good. The great thing about IoT is in its connectivity. I, or my family can communicate with a a connected device either through the phone (if not at home), but more conveniently by talking to a smart speaker. Common phrases around the house now include, “Google, play Comedians in Cars Getting Coffee on TV1” and “Google, turn on/off the kitchen lights”. I love my Chromecast and the easy smartphone interface makes it a joy. But now the interface is my voice, which makes things much more seamless. Another fun thing we like to do in our home now, although I’m not sure it warrants the hefty price tag is to fool around with the colors of the lights. For the 4th of July (same for Bastille Day!) we had red, white and blue in our kitchen. And for Halloween, we’ll dress up the lights again. We can also control these lights from afar.

The bad. Connectivity can be a funny thing. And when I say funny, I mean frustrating. Every morning I go downstairs to find one of my 3 kitchen smart lights on. I think this is more to do with power surges in my home, but still, it is a frustrating experience you don’t get with non-smart technologies where on/off switches control the power.  Another very, and I mean very, frustrating experience is having to repeat myself with my smart speaker. In some cases the speaker doesn’t hear me, or can’t hear me when I’m rocking out to louder music. In other cases, its NLP processor has a hard time differentiating between turn kitchen 1, 2 and 3 lights on, versus turn kitchen 1 light on, kitchen 2 light on and kitchen 3 light on.An annoying respnse is “Sorry, I don’t know how to help with that right now.” It doesn’t help that I’ve been told that I mumble.

The scary. Finally, there is a lot of concern surrounding IoT and  security. An interesting article by the NYT (https://www.nytimes.com/2018/06/23/technology/smart-home-devices-domestic-abuse.html) focused on how smart devices can be controlled for nefarious purposes, but there are so many more areas for abuse, from hackers, government eavesdroppers and even the smart speaker manufacturers themselves. I couldn’t say, or I won’t say, what I song I was singing at home, but Google Assistant interrupted and told me that it liked the sound of that. This was an unprompted response to a conversation were weren’t having. So they are listening, but who and for what purposes remains a mystery to me…

Read More