ACM’s Special Interest Groups seem pretty important, so they should be a good
source of current ideas. The annual
Hypertext and Social Media Conference
(HT)
should be a good supplement to WWW.
Browsing HT ‘21 reveals that it has more web-related topics, but the number of
submissions is much lower than that of WWW ‘21.
Journals note that prediction services can still make accurate predictions using a fraction of the data collected from a user device. They propose Cloak, which suppresses non-pertinent features (i.e. those features which can consistently tolerate addition of noise without degrading utility) to the prediction task. Cloak has a provable degree of privacy, and unlike cryptographic techniques, does not degrade prediction latency. Using the training data, labels, a pre-trained model and a privacy-utility knob, they (1) find the pertinent features through perturbation training, and (2) learn utility-preserving constant values for suppressing the non-pertinent data....
Meta 📑 Instead of changing the data or learners in multiple ways and then see if fairness improves, postulate that the root causes of bias are the prior decisions that generated the training data. These affect (a) what data was selected, and (b) the labels assigned to the examples. They propose the \(\text{Fair-SMOTE}\) (Fair Synthetic Minority Over Sampling Technique) algorithm which (1) removes biased labels (via situation testing: if the model’s prediction for a data point changes once all of the data points' protected attributes are flipped, then that label is biased and the data point is discarded), and (2) rebalances internal distributions such that based on a protected attribute, examples are equal in both positive and negative classes....
One of the grand dreams of the web is bringing people together so that they may find relevant information.
Sometimes this is implicit, e.g. Google Search’s PageRank learns from inter-page links which are mostly made by people, but there may be room for explicit collaboration when navigating the web. This section explores such explicit attempts.
#lost-in-hyperspace
A Group Asynchronous Browsing (GAB) Server Premise: Users submit their subject hierarchies (bookmarks) to the GAB server....
Link Prediction Given network-structured data, predict whether a link exists between two nodes.
General types of prediction tasks on graphs: graph-level (e.g. will a molecule bind to a receptor implicated in a disease?), node-level (e.g. what is the identity of each node?), and edge-level (e.g. does this edge exist; what value does it have?).
Applications include: predicting drug-drug interactions (common in treating patients with complex/co-existing diseases) as they may cause changes in the drugs' pharmacological activity ....
Political Ads Cambridge Analytica (CA) paid people to take in-app survey; mined FB profile data including friends' data; crafted tailored sensitive ads to sway-able voters. Elections are about emotions, not facts. Data science and social media can help us make sense of and manipulate the chaos.
An alternative argument. Political misinformation is:
Weak in high profile partisan races because pre-existing beliefs hardly change Strong when people don’t have string pre-existing opinions, e....
WWW ‘21: The Web Conference 2021 REST: Relational Event-Driven Stock Trend Forecasting REST, an event-driven stock trend forecasting framework, that overcomes two limitations of existing event-driven models. Models the stock context, and learns the effect of event information on the stocks under different contexts. Constructs a stock graph and designs a new propagation layer to propagate the effect of event information from related stocks. #stock-trend-forecasting #computational-finance
#graph-based-learning
The value of stock trend forecasting is not unanimous, e....
ACM’s Special Interest Groups seem pretty important, so they should be a good source of current ideas. The annual Hypertext and Social Media Conference (HT) should be a good supplement to WWW. Browsing HT ‘21 reveals that it has more web-related topics, but the number of submissions is much lower than that of WWW ‘21.