(Analysis by Nick Wu)

Unlike the silver version, we can't do a BFS for every query, as that would take $O(NQ)$ time.

Unlike other problems, where the queries have to be done online and one at a time, all of the queries are given to us at once and we just have to answer all of them within the time limit. This means we can order the queries in a certain order that make it easier for us to answer them offline.

In particular, we can read in the whole graph and queries, and then sort the queries in decreasing order of relevance threshold.

How does sorting them by decreasing order help us? If we think about the original BFS solution, we ignored edges if the weight of the edges was below the threshold. If we start with an empty graph and process queries, we can use all the edges from the previous query and add in any new edges that are now at least as large as the threshold we're querying against.

Note however, that now we're simply counting the number of vertices in a connected component. We can use a union-find data structure to maintain the size of every connected component and merge two components whenever an edge becomes valid.

import java.io.*;
import java.util.*;
public class mootube {
	public static void main(String[] args) throws IOException {
		BufferedReader br = new BufferedReader(new FileReader("mootube.in"));
		PrintWriter pw = new PrintWriter(new BufferedWriter(new FileWriter("mootube.out")));
		StringTokenizer st = new StringTokenizer(br.readLine());
		int n = Integer.parseInt(st.nextToken());
		int q = Integer.parseInt(st.nextToken());
		Edge[] edges = new Edge[n-1];
		for(int i = 0; i < edges.length; i++) {
			st = new StringTokenizer(br.readLine());
			int x = Integer.parseInt(st.nextToken())-1;
			int y = Integer.parseInt(st.nextToken())-1;
			int w = Integer.parseInt(st.nextToken());
			edges[i] = new Edge(x, y, w);
		}
		Arrays.sort(edges);
		par = new int[n];
		sz = new int[n];
		for(int i = 0; i < n; i++) {
			par[i] = i;
			sz[i] = 1;
		}
		Query[] queries = new Query[q];
		for(int query = 0; query < q; query++) {
			st = new StringTokenizer(br.readLine());
			int threshold = Integer.parseInt(st.nextToken());
			int start = Integer.parseInt(st.nextToken())-1;
			queries[query] = new Query(start, threshold, query);
		}
		Arrays.sort(queries);
		int[] ret = new int[q];
		int idx = 0;
		for(Query query: queries) {
			while(idx < edges.length && edges[idx].w >= query.w) {
				merge(edges[idx].a, edges[idx].b);
				idx++;
			}
			ret[query.i] = sizeOf(query.v)-1;
		}
		for(int out: ret) {
			pw.println(out);
		}
		pw.close();
	}
	
	static int[] par;
	static int[] sz;
	public static int sizeOf(int x) {
		return sz[find(x)];
	}
	public static int find(int x) {
		return par[x] == x ? x : (par[x] = find(par[x]));
	}
	public static void merge(int x, int y) {
		int fx = find(x);
		int fy = find(y);
		sz[fy] += sz[fx];
		par[fx] = fy;
	}
	
	static class Edge implements Comparable<Edge> {
		public int a, b, w;
		public Edge(int x, int y, int z) {
			a=x;
			b=y;
			w=z;
		}
		public int compareTo(Edge e) {
			return e.w - w;
		}
	}
		
	static class Query implements Comparable<Query> {
		public int v, w, i;
		public Query(int x, int y, int z) {
			v=x;
			w=y;
			i=z;
		}
		public int compareTo(Query q) {
			return q.w - w;
		}
	}
	
}