i don’t understand how calculate points of “Lottery” and “win”

Sorry for the delay in replying.

In this example, the point values (3 for the word ‘lottery’ and 2 for the word ‘win’) are just assumed. In a real NN, these values would be learned so that the predictions have the best chance of being correct.

Each row of the table represents one message. The numbers in each column are the number of times those words appear in that message.

For this particular algorithm, you count up the point values for each of these key words in each message.

So in the 2nd row, you have the word ‘lottery’ twice, for 2*3 = 6 points, and the word ‘win’ once for 2 points, so the sum is 6 + 2 = 8.

The example on the slide “Win, win the lottery!” is from the last row in the table (which you did not show in your image).