Visual Analytics for Eth2

Transcript

hello everyone this is simin from fudan university it's honored to be here to share some research in visual analytics for eth and i think everyone knows visualization and i want to share something from uh ideas of my own that we believe that one picture explores thousands of words you can see from this very classical example showing the napoleon's troops going further east and back to the west and with the graphics and web development we are more using a data driven approach and use a center design to design the visualizations and also

with the time passed in this recent 10 years visual analytics is proposed so what is visual analytics it is combination between of the automatic algorithms interactions the visual designs for enabling human in this analysis loop to understand the data so let's reach analytics there are many disciplines like we can apply with genetics today what i want to show is two explorer cases for the visual networks for eth the first one is the situation awareness for state of epochs and validators so one the analysis goal is that we want to understand the state of efforts and

then we want to understand attestation and committees including slot votes with this understanding we want to draw the conclusion relationship between sloth with problems and corresponding variations that's our analysis task and then with this task in mind we get this data with the information of participants information blocks and the information of the interaction between eth1 and h2 we focus on the status of epochs with this understanding we shall redesign these visualizations first one is the overview is this matrix view showing the

situation of the airbox aggregation of the airports each rectangle has 31 epochs and we can see the color indicates this number of the program matic epochs so and the horizontal rows is continuous and then with the mouse over we can see the details also we can click specific application detail and drill down to the second view second view showing the individual epoch that from the aggregated epochs visualization and here is block showing this one epoch and there are also continuous slot uh put there and there are three types

first one is the proposed the second one skip then first one is forked again we can still uh click to see the detail of the epoch of the interest go to the third detail field that's showing from the one airport and we have two colors indicates the attestation situation the first one is the green one is showing the normal one and the pink one showing the abnormal ones so with this we can understand the distribution of this epoch of the test statistician and also we show this detail will be showing these uh

problems of these participants in one epoch the color depth indicates the probability that certification is included and also the thickness of this line shows the connection indicates the length of the time used by the verifier from eligible to active and with the outside ring it shows the peak values of the number of problem slots that in which validator participates so we can see if his participants participate many problem slots so probably he has some problems so with this understanding we can draw some help the people get the situation awareness of this

epics and the validators so to have a short summary of the first one we provide an overview to detail views and also to in the middle view which shows the aggregate pattern from the overview we show is the epochs aggregation and by clicking each one you can go through the epoch details distribution and again by clicking each one of the epoch we can see the committee test citations and also the validatory situation so this visualization we provide list from overview to details for the users to understand the whole process the process and the status of networks and validators these are not for user first work

and the second work i want to share with you is the visual analytics for phishing scheme activity signal transaction data in dhh and this data is from the ether scan also we downloaded from xblock and the layout 3360 nodes with label fishing nodes and also 1000 and seven hundred normal nodes and also we have the all the transactions from these all the nodes so the data increases time step from two and the value it's basically is a graph data so we process this data and construct a graph and in processing time we train this genome model and then to understand

this node distribution node classification based on the label data and then we build up this visual analytic system how to see this complex visual analysis system so there are four components the first one is the timeline showing all the transaction numbers in each time period and second one is the glm view which shows the projection based on this high dimensional information derived from this gn basically we can understand this is the high dimensional feature of each node and the third one is the topological view shows the connection of these nodes each links means there is a transaction between these two nodes and the color indicates the selection

also the red ones indicate the nodes which labels in phishing nodes the last view is the dynamic network view which shows the dynamic process of this involving process of the network the transaction network we can see in different time there will be small transactions a large period and then get another small ones like this so let's see how we do the visual analytics in system so there's some demos that we record you can see this all these views are linked so in the left part it's a gm projection view we can see the

we provide a brushing function to help people understand other views like this topology view so we see when we select this range and it shows these nodes like english topology relationship so this noting the high dimensional is similar because we see this like all connected in the one big nose and these are all nodes without other nodes except this one into the center ones also in the time period we can see the peak transaction there's also some there's some times without making many transactions we can brush the time and as this as a filter to update all these views from also from the

gm and also topology view so you can see this updated and there's a force directory layout they are doing this layout process and after this layout is done you can see there are some patterns of these two big bridge nodes and also there are some labeled sufficient nodes are here with the ruling function we can see the detail of their block address so with this illustration of this linked view we can see the exploration process from the multiple perspective like from high dimensional data high dimensional perspective the topological process uh

perspective and the temporal process the perspectives and then we can drill down to the details of this dynamic process so okay there's also the detailed information that we can show that when you click each note it shows the basic information there also we can trace trace lists labeled ones notes and to see what happens from this this nodes to others we can see there's one level neighbor and the second level the network level is here so which means it goes from the other nodes and to the second one it also shows the statistic statistics

of this node's behavior and moreover we can show in the temporal dynamic network view to see the dynamic process and identify this there are three labeled efficient nodes they are all connected and into some normal nodes such as these findings so basically which i show you these two visual analytics approach for eth and also we have more different visualizations in different domains so interesting can come to visit my webpage and

thank you very much