Pre-requisite:

1. Set s to store linked list nodes.
2. Two head pointers pointing to the Y('T') shaped linked list.

Algorithm:

1.  Declare set 's' to store nodes of the linked list;
2.  Declare and define two pointer to node type variable 
    Node* temp1=head1,*temp2=head2 //for traversing

3.  While(temp1){ //until the first linked list is traversed
        Insert all the nodes (pointer to nodes of course) to the set; 
        //remember nodes to be entered not the data value
        Insert(s, temp1);
        temp1=temp1->next;
    END While
4.  While(temp2)
        IF (if temp2 is not present in the set) //not part of merged tail
            temp2=temp2->next;	
        Else
            return temp2->data; //first intersection point found
        End IF-ELSE
    End While

Note:

The nodes (pointer to nodes) are to be inserted into the set, not the node data. Let's say for an input example both heads have the same input data (for example say 1), but nodes are actually different (no intersection there). But if we maintain the set with node data, then that will return the 1, which is the wrong answer.

    1 → 2 → 3 → 6 ← 4 ← 1
                ↓
                8
                ↓
               NULL

For the above example, answer will be 1 if we start inserting node->data to the set instead of node. But the actual answer is 6.

Explanation with example:

    The above algorithm is pretty simple and self-explanatory.
    Discussion for above example:
    After processing the first list the set contains:
    Pointer to Node having value 1
    Pointer Node having value 2
    Pointer to Node having value 3
    Pointer Node having value 6
    Pointer Node having value 8
    NULL
    When second list is processed and checked,
    Pointer to node having value 1 is not in set (any doubt!Why? think yourself)
    Pointer to node having value 4 is not in set (No doubt)
    Pointer to node having value 6 is in the set 
    (then what’s the difference between the first case where you have doubt!!! 
    Reason is different address of the nodes though same value since those are 
    different node, but in the second case the common node is same node)
    Thus it returns 6.

C++ implementation:

#include <bits/stdc++.h>
using namespace std;

class Node{
	public:
	int data; // data field
	Node *next;//next pointer
};

Node* creatnode(int d){
	Node* temp=(Node*)malloc(sizeof(Node));
	temp->data=d;
	temp->next=NULL;
	return temp;
}

int intersectPoint(Node* head1, Node* head2)
{
	set<Node*> s;
	Node* temp1=head1,*temp2=head2;
	
	while(temp1){
		s.insert(temp1);
		temp1=temp1->next;
	}
	while(temp2){
		if(s.find(temp2)==s.end()){
			temp2=temp2->next;
		}
		else{
			return temp2->data;
		}
	}
	return -1;//not at all a Y-shaped one
}

int main(){
	cout<<"Linked list built like modified example\n";
	
	Node* head1=creatnode(1);
	Node* head2=creatnode(1);
	
	head1->next=creatnode(2);
	head2->next=creatnode(4);
	head1->next->next=creatnode(3);
	head1->next->next->next=creatnode(6);
	head2->next->next=head1->next->next->next;//merged
	head1->next->next->next->next=creatnode(8);

	if(intersectPoint(head1,head2)==-1)
		cout<<"Not Y/T shapped at all";
	else{
		cout<<"intersection point is:";
		cout<<intersectPoint(head1,head2)<<endl;
	}
	
	return 0;
}

Output

Linked list built like modified example
intersection point is:6