Writing reviews is important but sometimes repetitive and time-consuming. Hence, today I built a tool to help automatize the process of review writing. You can try it at the website below:
This tool let you select some items and this will add predefined sentences to your review. Of course, this tool is not supposed to replace a human and generated reviews should be viewed as draft and edited by hand to add more details.
If you like the tool, you may boomark it and share it. And if you would like more features, please let me know. For example, if you would like that I add more content to the tool, please leave a comment below or send me an email.
** Credit ** That project is a modification of AutoReject (https://autoreject.org/)by Andreas Zeller, which was designed as a joke to automatically reject papers. I have reused the template but modified the text content to turn it into a serious tool. #review#academia#reviewgenerator#reviewprocess#journal#conference
For over two years, I have worked as associate editor-in-chief for a major journal published by Springer. For this job, I have handled around 500 papers from all the steps of the review process. Recently, I have decided to stop this work as it was taking too much of my time and as time is limited in life, I wanted to give priority to other aspects of my life. I was very happy to be associate editor-in-chief for that big journal as this has allowed me to learn many things. I will discuss what I have learned in this blog post. The key things that I have learned are:
Being an editor can be a lot of work. For a big journal, the workload of an editor is quite high. I would spend about one hour every day to try to find reviewers for papers, reading the papers or do other related tasks. One hour may not seem a lot but when you have a research team, a family and other things in your life, it is a lot of time and you have to think about your priorities. With this 1 hour, I could instead do some sport, spend more time with my kid or spend more time on my own research.
Finding reviewers for a paper is not so easy. Authors of research papers often think that reviewers are easy to find. But the truth is that often it is a hard task to find reviewers. The editor has to invite numerous reviewers. Each reviewer has one week or more to accept the invitation. And often the potential reviewers would not answer or decline. Thus, it can happen that it takes over one month to find suitable reviewers for a paper. Generally, if a topic is unpopular, it can be quite hard to find reviewers.
Reviewers are often late. Many authors do not know that it is quite common that reviewers submit their reviews late. Often, it is by a few days, but some reviewers can be late by up to a month in some cases. When a reviewer is late, the journal will automatically send reminders to the reviewer but still some reviewers can receive dozens of reminders and ignore them. In this case, the editor has to step in and find more reviewer(s).
The work of an editor can be repetitive. Although being an editor is interesting and it is also very important for academia, I feel that some work is repetitive such as clicking to find reviewers. Other people will have different opinions but rather than doing such tasks, I would rather spend time working on my own research.
When submitting a paper, if you can enter keywords or select topics from a taxonomy to describe your paper, choose them wisely. It is important to choose keywords that really describe your paper well as this will be used to find reviewers for your paper and can accelerate the review process.
Some reviewers are unethical. As editor, it is important to read carefully what the reviewers write as there are several reviewers that would act unethically such as by asking authors to cite many of their papers to increase their citation count. When i would see this, I would intervene to stop this, of course, but there are always a few reviewers that are trying to do this.
If authors do not do enough effort to revise their paper well (after a round of review), it is not uncommon that a reviewer will change his mind and suggest to reject the paper in the next round. The reviewers are doing their work for free. Some reviewers are very patient and will accept to review the same paper multiple times. But others will not be patient if the author does not make sufficient effort to address issues in the paper. Thus authors should always do their best to solve the problems in their paper.
That was just a short blog post to talk a little bit about the work of an editor and the review process of a journal. I hope it does not sound negative because actually, I have learned many things from this work and I do not regret accepting to do this work.
Hope that this has been interesting. If you have any comments or questions, please leave them in the comment section below!
Special Session on Data Science for Social and Behavioral Analytics
This is to let you know that I co-organize a special session called DSSBA 2022 at the IEEE DSAA 2022 conference, and we need your papers ;-).
This special session is about social and behavior analytics. This includes topics such as:
• Efficient and scalable algorithms for behavioral and social analytics • Evaluation of behavioral analytic models • Social computing and behavior analytics • Interpretation of data science models under cognitive theories • Business process analysis • Customer behavior analysis • Social network analysis • Abnormal behavior detection • Group behavior analysis • Behavior prediction • Models for targeted and personalized behavior services • Privacy-preserving behavior and social analytics • Security models and algorithms for behavioral analytic • Behavioral economics with data analysis • Case studies and real-world applications of the above
All the accepted papers will be published as regular papers in the proceedings of IEEE DSAA 2022. The papers are published by IEEE and EI-indexed.
The deadline is: 1st June 2022 The format is up to 10 pages in IEEE double-column format.
Today, I post one more new video to explain concepts about pattern mining. In the new video， I talk about high utility itemset mining, and explain the HUI-Miner and FHM algorithms. Those are two popular high utility itemset mining algorithms that have been used in hundreds of research papers.
I have posted a new video about pattern mining, explaining the PrefixSpan algorithm. It assumes that you know already what is sequential pattern mining. If you are not familiar with sequential pattern mining, you can first watch my video Introduction to sequential pattern mining.
Today, I will reveal some upcoming features of SPMF, which I am currently testing and will be released in the next version (2.54), probably in about 1 week.
1) The first new feature is that it will be possible to launch an algorithm in a separated virtual machine instead of using a thread in the same virtual machine. This option is useful for running performance experiments as it ensures that a new virtual machine is used for each algorithm execution. This avoids the problem that the memory is not released by the Java Garbage collector.
2) The second new feature is a time limit. It will now be possible to set a maximum time limit and to kill an algorithm automatically after exceeding that time limit.
The features are already implemented and I have tested them on Windows. I will be testing them on Linux later to see if it also works before releasing them.
If you have any suggestions for other features, you may leave a comment in the comment section below. I am happy to receive any comments to improve the software 🙂
This is a short blog post where I will explain how to call SPMF as an external program from another Java program to execute an algorithm.
Before we start, it should be said that there are multiple ways to use SPMF. It can be used as a standalone program with a graphical user interface and from the command line. Moreover, the code of SPMF can be directly integrated in other Java programs, and SPMF can also be called using unofficial wrappers from other languages such as Python and R.
If you want to use SPMF from a Java program, it can be desirable in some cases to call SPMF using its command line interface rather than integrating the code of SPMF directly in your Java program. The adavantage is that SPMF is then executed as a separated process on your computer and it may be easier to maintain. How to do this?
First, you should download spmf.jar from the download page of the SPMF website and put it in the same folder as your Java program.
Second, you should lookup which algorithm you want to use in the documentation webpage of the SPMF website. For this example, lets say that we want to call the Apriori algorithm on a file called “contextPasquier99.txt” with the parameter minsup = 40% and save the result in a file “output.txt“.
According to the documentation of Apriori, we should write a command like this to execute it from the command line:
java -jar spmf.jar run Apriori contextPasquier99.txt output.txt 40%
From Java code, we can write do like this:
List commandWithParameters = new ArrayList();
commandWithParameters.add("java");
commandWithParameters.add("-jar");
commandWithParameters.add("spmf.jar");
commandWithParameters.add("run");
commandWithParameters.add("Apriori"); // Algorithm name
commandWithParameters.add("contextPasquier99.txt"); // input file
commandWithParameters.add("output.txt"); // output file
commandWithParameters.add("0.5%"); // parameter
ProcessBuilder pb = new ProcessBuilder(commandWithParameters);
pb.redirectOutput(Redirect.INHERIT); // This will redirect the output of SPMF to the console
Process process = pb.start(); // Run SPMF in a separated process
Running this code, will execute the Apriori algorithm and write some statistics in the console:
============= APRIORI - STATS =============
Candidates count : 11
The algorithm stopped at size 4
Frequent itemsets count : 9
Maximum memory usage : 7.322685241699219 mb
Total time ~ 0 ms
===================================================
If you want your Java program to wait for the completion of SPMF before continuing, you can add this line:
int exitValue = process.waitFor();
It is also possible to stop the process using this:
If you like these videos, you can also check the video page of the SPMF software website, where I have put more videos to explain algorithms. Also, you can check out my Youtube channel, which contains also various recorded talks that I have given.
I you would like that I make a video about a particular topic, you may leave a comment below, and I will consider it if I have time and I think the topic is good.
The TKU-CE algorithm for heuristically mining the top-k high-utility itemsets with cross-entropy (thanks to Wei Song, Lu Liu, Chuanlong Zheng et al., for the original code)
The TKU-CE+ algorithm for heuristically mining the top-k high-utility itemsets with cross-entropy with optimizations (thanks to Wei Song, Lu Liu, Chuanlong Zheng et al., for the original code)
The TKQ algorithm for mining the top-k quantitative high utility itemsets (thanks to Nouioua, M. et al., for the original code)
Besides, since December, four more algorithms have been released (in SPMF 2.50 and 2.51):
The SFU-CE algorithm for mining skyline frequent high utility itemsets using the cross-entropy method (thanks to Wei Song, Chuanlong Zheng et al., for the original code)
The POERMH algorithm for mining partially ordered episode rules in a sequence of events, using the head support (thanks to Yangming Chen et al. for the original code)
The SFUI_UF algorithm for mining skyline utility itemsets using utility filtering (thanks to Wei Song, Chuanlong Zheng et al., for the original code)
The HAUIM-GMU algorithm for mining high average utility itemsets (thanks to Wei Song, Lu Liu, et al. for the original code)
Need your contributions!
For the SPMF project, we are always looking for new contributors. If you are interested to participate (e.g. contributing code of new algorithms, bug fixes, etc.), you can contact with me at philfv AT qq.com.
Typhoons can be very destructive. Predicting their paths is important to be prepared when they arrive. In this blog post, I will talk briefly about an applied research topic which is to predict the paths of typhoons. This blog post is based on a recent research paper published in Neural Computing and Applications, where I have participated as co-author:
Xu, G., Xian, D., Fournier-Viger, P., Li, X., Ye, Y., Hu, X. (2022). AM-ConvGRU: A Spatio-Temporal Model for Typhoon Path Prediction. Neural Computing and Applications, Springer,
Over the years many models have been proposed for typhoon path prediction. But the accuracy of these models could be improved. In general, we want to have models that are as accurate as possible.
Predicting typhoon paths is a difficult problem because it involves spatial data and temporal data, that is described using numerous features. Moreover, some features are 2D features while others are 3D features and combining them is also a challenge.
To address this issue, in the above paper, we presented a deep learning framework to perform accurate predictions of the paths of typhoons. The model is called Attention-based Multi ConvGRU (AM-ConvGRU).
For that research project, typhoons data was obtained from two sources: (1) the China Meteorological Administration (CMA) and (2) the European Centre for Medium-Range Weather Forecasts (ECMWF). The first provides data about 2D typhoons while the second provides 3D typhoon data. The data covers typhoons in the Western North Pacific (WNP) basin. Here is a visualization of typhoon paths from the paper:
After obtaining the data, the data has to be preprocessed. In particular, the 2D typhoon data is transformed into 53 features, according to a method called CLIPPER. These features are depicted in the table below as example.
Similarly, the 3D typhoon data has to been prepared. This is done by dividing the earth into a grid of 1 degree by 1 degree, by geopotential, and then looking more closely at the zone around the typhoon center. I will skip the details. But the result is a 3D time series structure:
After that, the deep learning model is trained using the 3d and 2D typhoon data. This is an overview of the model’s architecture:
I will skip the details.
To evaluate the proposed model, it was compared with state-of-the-art models. It was shown that the proposed model can generally provide better predictions.
To show a little bit more clearly what is the output, here is an illustration of some prediction by the proposed model, a baseline model, and to the historical path for Typhoon Mangkhut:
It can be seen that the proposed model is closer to the historical path than the baseline by over 50 km. Here is another example for Typhoon Talim:
The improvement of distance error for the proposed model over the baseline is over100 km.
Hope this has been interesting. This is just a very short overview of the topic of typhoon path prediction. If you are interested, please check the paper!
