FIELD OF THE INVENTION
This invention relates to the field of education, and more particularly this invention elaborates
the AI & Block chain mechanism.
BACKGROUND OF THE INVENTION
Technology is the initial layer of digital transformation. AI and ML-based technologies have
expanded in capabilities and accessibility in recent years. By recognising AI's future benefits
and drawbacks, humanity can benefit from AI. AI research and legislation aim to balance
societal security with potential dangers and obstruction. Indian government aims include AI
development, adoption, and promotion. The preciseness and verisimilitude of the details
regarding where crimes occur and the depiction of crimes gave a way to understand such
crimes in other nations. McGuire and Holt's also high lights the remarkable and much-needed
Routledge Handbook of Technology, Crime and Justice, which shows criminology's growing
technological interest. The most significant aspect of this research is updating judges to be
computer specialists; all judges should be educated to use this technology. Using AI, John
McCarthy's major developing technology that sees data as it is. Humans must pick data
before reasoning, which can lead to errors. COPLINK is a diverse and effective application
of information technology. COPLINK is a licenced software that helps police personnel assist
the community by conducting research and solving real-world crimes. This COPLINK
project combines the University of Arizona's AI Lab with the Tucson Police Department's
law enforcement . It's important to consider the many ethical difficulties in the criminal
justice system, which requires moral judgements and right and wrong decisions. Data mining
helps create criminal detection models. Since the model is unbiased and accurate, ethics have
been maintained.
SUMMARY OF THE INVENTION
Through the study above, we can conclude that by trial and error with various algorithms, we
could draw some crucial points with the help of the Logistic Regression Algorithm. This
research would surely help the law enforcement agencies understand the root cause of the
crime as if there was any political movement, natural crisis, or else massive dropouts in the
particular state which led to a person committing the crime. As we cannot rely on this model
completely in sentencing the accused, his/her parenting, upbringing, society, and teachings
should also be gone through to understand the reason behind committing the crime, as we all
know the law enforcement agencies or government can only bestow law upon us. Still, the
root cause of this crime should be found and eradicated. The various sectors can benefit from
this new technology provided that it is not used for somebody’s harm for it to behave in
unpredicted and potentially harmful ways. Thus, the proper judicial monitoring of data fed
can enjoy this model’s beauty.

BRIEF DESCRIPTION OF THE INVENTION
The most crucial concept for approaching this topic would be the understanding of
recidivism; through this model, we can keep a close watch on the behavior of various states
and the crime committed. As shown in the Fig. (1) we can see howthrough certain steps our
data is being processed in order to get the desired results. Different programming languages
and environments enable ML research and de-velopment of its application. Python language
has a tremendous growth within the scientific computing communities in the last decade, so
in this case most re- cent ML and deep learning libraries are associated with Python based.
Python is used to prepare the model of predictive analysis and us- ing the EDA Exploratory
Data Analysis Then a particular data becomes large orwe need to understand some complex
relationships in the variables. Through this we can perform the molding of such data for
better investigation purpose. First, the data is loaded in python and then we perform data
cleaning and exploring the information in the variables. Pandas which provide data frames
are imported using python, Matplotlib provides plotting support, and Numpy provides
scientific computing within dimensional object support as seen in Fig.2 . Secondly,
standardization and visualization of data is very important to ensure that data fits the
assumptions of the models. The Universal Rule of Law states that human rights, democracy
and development depend on the level of progress the organizations and governments can
achieve on the criminal justice front.
The primary and crucial objectives of the criminal justice are controlling and preventing
crime, maintaining law and order, protecting fundamental rights of victims along with the
people in conflict with law, punishment and rehabilitation of those adjudged guilty of
committing of crimes, and protection of life and property against crime and criminality in
general. It is considered to be the primary obligation of the state under the constitution of
India. This would thus give an overview how every police station can update their data and
predict the criminal behavior of the crime or any data available. Importing various libraries
and functions is the positive point of using python in this research since the data could be
easily adjusted, it can be seen in Fig. 3 and 4. Accurately predicting rare events is difficult, so
the probability of having themin data is low, and the probability of training the algorithm is
also low. Therefore, we only need a few percentages of the event to be able to train, to
ensure thatwe have a reasonable chance to define how correctly a person or state is likely to
develop the behavior or motive of committing a crime. Importing pandas willlet us easily
search the columns by name and see how many times this is true.
Also, in the last column seen in the Fig.3 threat columns are mentioned which is
categorically divided into binary 1s and 0s where 1s define that the attacks are increasing
drastically whereas 0s define that the motives are mild. When a crime is predicted there will
be questions arise regarding how an algorithm or code can be trustworthy. This research
would, therefore, throw light on this area where the data itself would be deciding
everything, the more real the data the more effective the accuracy would be. Data mining and
predictive analysis play an essencetail role in our life. Now if we look into the data available
very carefully, we can find whichever states having high unemployment rate according to
report by theCentre for Monitoring Indian Economy. It is note worthy, that such states have
high cybercrime rates which further denotes that in various states computer is used as a
source to dupe money through various online frauds. The reason behindthis is maintaining the
anonymity and causing the harm because of vengeance or other motives. Cybercriminals
mostly exploit the high-speed internet available at a lower cost to commit various criminal
activities without being caught unless the states possess properly well-maintained cyber
security labs to curb such crimes. The CMIE report further reveals that people belonging to
age group 40 to 59 years have been successfully able to retain their jobs whereas people aged
below 40 years were expelled out of their respective jobs which lead to social tension, desire
of revenge, anger and other motives to launch such cyber-attack. The data shown in Fig. (3)
presents the topmost cyber-crimes happened in various states of India until 2019. So far,
which includes such crimes as bullying on social media and not full-fledged crimes wherein a
lot of technical skills are required, this shows that certain age groups of people have launched
such attacksto malign the image of the victim In Fig. 4 we can see the features of data, a
feature is something that’s used to determine a result, and a column is a physical structure
that stores the value ofa feature or a result.
In Fig. 12, using shape function the data is displayed in the format of rows and columns; here
we have 36 rows and 13 columns; also, we check whether there are any null values present in
the data sets shown in Fig. 12. Matplot library is used to create a function that cross plots
feature so that we can see when they are correlated. Data is then inspected in order to
eliminate any additional columns or rows to with no values that we no longer required. The
duplicates including the same values are removed the same way. This is done to arrange our
data since visual inspection may be error-prone and cannot deal with the critical issue of
correlated columns. Thus, pandas help in understanding such null values and therefore
identifying it in our data as we can see in Fig. 6, Is Null method willcheck each value on the
data frames for null values. Similarly, Matplot library is used to create a function plots
features so that we can see when the data is cor- related: the color in yellow denotes the very
positive correlation as seen in Fig. 11 and other color denotes that the data is not well
correlated. In Fig. 11 we can see that column names on the horizontal and vertical axes is a
matrix showing whichcolumn contains the data that are correlated with values.
In [3]: os.getcwd()
os.chdir (‘C:/Users/Puneet/CRIME RECORD’)
os.getcwd()
Out[3]: ‘C:\\Users\\Puneet\\CRIME RECORD’
In [20]: Cyber_data = pd.read_csv(‘Cyber new.csv’) # read dataset
Cyber_data.head()
Out[20]:
As we can see in Fig. (4) and (5), data is fetched from the file path and utilized for the
further data cleaning and correlating.
In [3]: os.getcwd()
os.chdir (‘C:/Users/Puneet/CRIME RECORD’)
os.getcwd()
Out[3]: ‘C:\\Users\\Puneet\\CRIME RECORD’
In [20]: Cyber_data = pd.read_csv(‘Cyber new.csv’) # read dataset
Cyber_data.head()
Out[20]:
MOLDING THE DATA
After cleaning the data of any extra columns or null values, we proceed to mold-ing the data by
inspecting if there are any issues. Algorithms are largely mathemat-ical models which work
best with numeric quantities and once the data molding is done, we can use this data for
further training the algorithm as seen in Fig. 6 count, mean, std, etc. is calculated so that the
data is molded accurately. Therefore,in machine learning, a lot of data manipulation is done
for trial and error and predicting the best of the accuracy. When the data is manipulated it’s
very easy tochange the meaning of the data what also helps in understanding if data has gone
wrong anywhere. The entire model is created in Jupyter Notebook, therefore keep-ing track of
all the changes and updates have been done automatically. We also have the interactivity of
the python interpreter usingwhich we can make our data simpler for the prediction, as seen in
Fig.6 and 7.
TESTING MODEL’S ACCURACY
In this section we will discuss the role of the Machine Learning algorithm. Analgorithm can
be defined as an engine that drives the entire process. For our prediction, we will use data
containing examples of the results and try to predict the future using the scikit learn and the
algorithm’s logic the data is analyzed. This analysis evaluates the data concerning a
mathematical model and logic associated the algorithm, and the algorithm then uses the
results of this analysis to adjust internal parameters to produce a model that has been trained
to best fit the features and give the best results. The best result is defined by evaluating a
function specific to aparticular algorithm.
Therefore, the fit parameters are stored and hence the model is now trained. Further, we use this
model to predict on the real data. We use the Sci-kit learn package in python to predict on the
real data. The parameters of the trained model along with the python code is used to predict
whether the state is in threat of cyber-attack or no. Selecting an appropriate algorithm from
scikit learning was the toughest part which we faced while researching on this. Prediction
means supervised learning so eliminating all other algorithms was my main goal,
furthermore, prediction can be divided into two more categories regression and classification,
where regression means a continuous set of values. Predicting binary outcome whether the
threat is there or not; we further eliminated all the algorithms that do not support classification
in general and especially binary classification. Naïve Bayes, Logistic Regression and Decision
Tree are algo- rithms which support classic machine learning algorithms and also provide
excel-lent help in understanding more complex algorithms.
CHECKING THE ACCURACY
Since we know that through our research aimed to predict whether a particu- lar State/UT is
at a higher risk of cybercrime when using such variables as Per- sonal revenge, Anger, Fraud,
etc. In order to predict this relationship, we have used a statistical technique Logistic
Regression. Before we move to modeling, wehave to check if there is any correlation between
the independent variables, in other words, we have to check if there is a relationship between
the independent variables (example Personal revenge, Anger, Fraud etc.). In the correlation
plot, we should ignore the diagonal block as the diagonal block in yellow represents the
correlation with itself (i.e., Personal revenge and Personal revenge) in which we are not
interested. Yellow color represents high correlation, light green color represents moderate
correlation, dark green color represents low correlation, and complete dark color represents
no correlation. So, from the plot we can say that there is a high correlation between Sexual
exploitation and Anger, spreading pira-cy and prank etc. as if we see the block of these variables
in the plot, they are yellow in color. There is a moderate correlation between Spreading piracy
and Causing disrepute, Prank and Inciting hate against country etc. as if we see the block of
these variables in the plot, they are yellow in color. Similarly, we can say that the variables
with darker blocks have less of no correlation. We have divided the datainto X and Y where X
is the independent variable and y denotes the dependent variable. So are independent
variables being Personal revenge, Anger, Fraud, etc.and our dependent variable is risk.
Further checked the shape of X (just a sense check), then divided the variable into train
and test, (we will use the X_train and y_train to train the logistic re- gression model and
then test the model using X_test and y_test). Now we have used the function to build a
logistic regression model using the data X_train and y_train. We are using this logistic
regression when our dependent variable has dichotomous type, i.e., True/False,
Absent/Present etc. Now having built a model,we have predicted the expected values of y
using X_test. After predicting the ex-pected values for y we will now check the accuracy
of the model. The accuracy of the model depends on the number of cases we have
predicted correctly, i.e., the number of times we have predicted that the State/UT is at risk.
The state was actually at risk and the number of times we have predicted that the State/UT
is not at risk and the state was not at risk. As seen in Fig. 11, 12, and 13, we can see that
how the model behaves in predicting the accuracy of the threat in the states.

CLAIMS
1. The Universal Rule of Law states that human rights, democracy and development
depend on the level of progress the organizations and governments can achieve on the
criminal justice front.
2. Therefore, we only need a few percentages of the event to be able to train, to
ensure that we have a reasonable chance to define how correctly a person or state is
likely todevelop the behaviour or motive of committing a crime.
3. Prediction means supervised learning so eliminating all other algorithms was my main
goal, furthermore, prediction can be divided into two more categories regression and
classification, where regression means a continuous set of values.
4. Data is then inspected in order to eliminate any additional columns or rows to with no
values that we no longer required. The duplicates including the same values are
removed the same way.
5. Now having built a model, we have predicted the expected values of y using X_test.
After predicting the expected values for y we will now check the accuracy of the
model.
6. This analysisevaluates the data concerning a mathematical model and logic associated
the algorithm, and the algorithm then uses the results of this analysis to adjust internal
parameters to produce a model that has been trained to best fit the features and give
the best results.
7. There is a moderate correlation between Spreading piracy and causing disrepute,
Prank and Inciting hate against country etc. as if we see the block of these variables in
the plot, and they are yellow in colour.