SMS and Social Media Messages Classification Scheme for Disaster Response of Sorsogon Provincial Disaster Risk Reduction and Management Office (SPDRRMO)

For my final Immersion blog, I decided to discuss the content of my research required by the immersion program. While working in the SPDRRMO, I had a chat with one of the rescuers that facilitate Quick Response and Rescue Operations if there is a need for one in times of Calamities and Disaster.  I asked him how they were able to get the request for rescue and how they validate these requests.  They said that they treat every request as valid and they can receive these request for help through SMS normally or even Facebook.  I asked them about how they process these requests.  They told me that their primary problem is the bulk of request that they receive during calamities. These messages came all at the same time.  Some of them are valid requests, some are SPAM, some are usual messages that has nothing to do with rescue operations.  For the valid messages they categorize them as to Alert 1 (those requiring immediate attention), Alert 2 (those requiring attention but not immediate, possibly within the day), Alert 3 ( those requiring attention but not immediate, possibly within the week or more).  Upon analysis, I know already that the situation could be solved by a Data Mining Classification Algorithm, most probably Naive Bayes Algorithm.  But the method requires that there must be a collection of previous information with classification  which will serve as a basis for classifying new information.  These collection is called learned classifier. I was able to finish my research and decided to present it to them a week after presenting it to the faculty of TIP....

Below is the research title and the abstract of my completed research...

-Aris Ordonez


--------------------------------------------

RESEARCH TITLE

SMS and Social Media Messages Classification Scheme for Disaster Response of Sorsogon Provincial Disaster Risk Reduction and Management Office (SPDRRMO)

ABSTRACT

This study classifies the type of messages sent through Social Media or Mobile SMS to the Sorsogon Provincial Risk Reduction and Management Office (SPDRRMO)  using machine learning algorithm to allow facilitate response during disasters. The method utilized Naïve Bayes Classification algorithm, a self-learning algorithm that is capable of classifying text information based on probabilities derived from past experiences and rectified mistakes. It was made possible by employing Natural Language Processing technique that allowed the processing and conversion of messages based on natural languages into a machine-ready data format that may be used as input by the Naïve Bayesian Classification Algorithm.  The output is a classified message according to its pre-defined classification significant to quick response operations.

Comments

Post a Comment

Popular posts from this blog

Sorsogon Provincial Disaster Risk Reduction and Management Office

Image
-Aris Ordonez Sorsogon Province is the last province in the south side of Luzon Island. It is part of the Bicol Region which is prone to different forms of natural disasters.  The region is a constantly visited by super typhoons formed in the Pacific Ocean.  It is also surrounded by active volcanoes that from time to time bring devastation to the region.  Flash floods and landslides are also common calamities that bring havoc to the place. These natural phenomenon can no longer be prevented and eventually became part of the lives of the thousands of residents of Sorsogon.  Resiliency became inherent to all Sorsoganons. The Sorsogon Provincial Disaster Risk Reduction and Management Office is the authorized government agency of the province to provide needed response in times of calamities and disasters. The agency also has the mandate to create policies and innovation that may provide risk reduction it times of needs.  The management of such is also a...
Read more

The SPDRRMO and its SYSTEM

Image
During my second week of stay in the Sorsogon Provincial Disaster Risk and Reduction Office, I got an opportunity to talk with the head of the agency.  Engr. Renan Dimaano, a civil engineer by profession.  He is the head of the office and the one in-charge of running the agency in attaining its objective.  He showed me the current system being utilized by the office in handling disaster and rescue operations. For disaster management, the agency needs to monitor a lot of information in order to provide good intervention and decision making.  For the rescue operation, a good and reliable communication line must be available for proper dispatch and order dissemination. For a province consisting of 12 municipalities, this will not be an easy task.  Engr. Dimaano showed me the system that they are currently using.  Most of these are provided by the national government through NDRRMC and the PAGASA-DOST.  They utilized these systems in monitoring upcoming ...
Read more

Possible Dissertation Topic

Prior to my entry to the DIT program of the Technological Institute of the Philippines, I was instructed to submit a concept paper on possible topic or area that I intend to tackle in my dissertation after the academic courses of the DIT Program. I was then firm on going for emphatic computing in application to Disaster Management and Response.  This was possibly because of my previous experiences in our province which has been constantly visited by typhoons and other natural calamities.  I wished to make a difference then along this area.  My first research output in the DIT program is a simulation concept that would predict the possible tracks that lava would take during volcanic eruptions with considerations to other significant parameters associated with this natural phenomenon.  I read a lot of researches along this area and found out that the concept of lava flow simulation can be modified to handle other similar natural phenomenon like flood and tsunamis. ...
Read more