recommendation.c

C编写的用来实现search engine的推荐功能

/**
 * recommendation.c CGI program for outputing recommendations 
 * by zhiyong, louisville, Dec..10
 * $Id:  Exp $
 **/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <strings.h>

#include "recdef.h"
#include "session.h"
#include "content_rec.h"
#include "collaborative.h"
#include "markov.h"

static char *cookie_name = "WebMiningProxy";
static char markov_two_gdbm[256] = "/home/zhiyong/software/apache2/cgi-bin/markov_two.gdbm";
static char markov_three_gdbm[256] = "/home/zhiyong/software/apache2/cgi-bin/markov_three.gdbm";
static char site_root[256] = "http://webmining.spd.louisville.edu:8090";
static char libr_root[256] = "http://library.louisville.edu";


int main()
{
	int i,iRet;
	char* Cookie_ID = NULL;
	char sessionID[256] = {0};
	char* ptr = NULL;
	char* pptr = NULL;
	
	SESSION_LINKS session;
	REC_LINKS rec_cf, rec_mk;
	
	char buf[20481];
	
	int linkNum;
	int recNum1, recNum2;
		
	char curLink[256];
	char *q = NULL;
	char query[256];
	
	printf("Content-type: text/html\n\n");	

	bzero(sessionID, sizeof(sessionID));
	bzero(&session, sizeof(SESSION_LINKS));
	bzero(&rec_cf, sizeof(REC_LINKS));
	bzero(&rec_mk, sizeof(REC_LINKS));
	
	q = getenv("QUERY_STRING");
	if(q == NULL || strchr(q, '=') == NULL)
	{ //browsing mode
		Cookie_ID = getenv("HTTP_COOKIE");
		//printf("<H3>Your Cookie ID:%s</H3>", Cookie_ID);
		if(Cookie_ID != NULL)
		{
			ptr = strstr(Cookie_ID, cookie_name);
			if(ptr != NULL)
			{
				if(strlen(ptr) >=256) 
				{
					fprintf(stderr, "session ID length too long\n");
				}
				else
				{
					strcpy(sessionID, ptr+strlen(cookie_name)+1);
					pptr = strchr(sessionID, ';');
					if(pptr != NULL) *pptr = '\0';
				}
			}
		}
		else //new user
		{
			printf("<H3>Welcome New User (%s)</H3>", Cookie_ID);
			return 0;
		}
			
		linkNum = getSessionUrlsFromGdbm(sessionID, &session);
		if(linkNum < 0)
		{
			fprintf(stderr, "getSessionUrlsFromGdbm failed\n");
			return -1;
		}
		else if(linkNum == 0)
		{
			printf("<H3>Welcome New User (%s:%s)</H3>", Cookie_ID, sessionID);
			return 0;
		}
		
	}
	else
	{//testing mode plust query
		ptr = strchr(q, '=');
		if(ptr != NULL) pptr = ptr+1;
		if(pptr != NULL)
		{
			strncpy(query, pptr, 256);
			query[255] = '\0';
		}
		else
		{
			printf("<H3>Decoding session failed</H3>");
			return 0;
		}
		ptr = strchr(query, '+');
		if(ptr != NULL && ptr+1 != NULL)
		{
			*ptr = '\0';
			strcpy(session.urls[0].url, query);
			strcpy(session.urls[1].url, ptr+1);
			session.num = 2;
		}
		else
		{
			strcpy(session.urls[0].url, query);
			session.num = 1;
		}
		linkNum = session.num;
	}
	printf("<H3>Your visit history(%d):</H3>", session.num);
	for(i=0;i<linkNum;i++)
	{
		printf("<LI><a href=%s>%s</a></LI>", session.urls[i].url, session.urls[i].url);
	}
	//collaborative filtering results
	recNum1 = getCollaborativeFilteringRecs(session, &rec_cf);
	if(recNum1 < 0)
	{
		fprintf(stderr, "getCollaborativeFilteringRecs failed\n");
		return -1;
	}
	//printf("<H3>Profile-based Collaborative Filtering Recommendations(%d)<a href=http://webmining.spd.louisville.edu:8090/profile.txt>[reference profiles]</a></H3>", recNum1);
	for(i=0;i<recNum1;i++)
	{
		strcpy(curLink, rec_cf.urls[i].url);
		myStrToLower((unsigned char*)rec_cf.urls[i].url);
		myStrToLower((unsigned char*)curLink);
		printf("<LI><a href=%s%s>%s</a>(%.0f)</LI>", libr_root, curLink, curLink, rec_cf.urls[i].wt);
		if(i >=MAX_RECS_SHOW/2-1) break;
	}
	//printf("<H3>KNN-based Collaborative Filtering Recommendations </H3>");
	printf("<p>Under Construction</p>");
	//Markov model based recommendation
	recNum2 = getMarkovRecs(session, &rec_mk);
	if(recNum2 < 0) 
	{
		fprintf(stderr, "getMarkovRecs failed one\n");
		return -1;
	}
	if(recNum2 > 0)
	{
		printf("<H3>Markov-based Recommendations(%d)</H3>", recNum2);
	}
	else
	{
		printf("<H3>That's Strange No Markov Based Recommendations</H3>");
	}
	printf("<!--Markov_Begin-->");
	for(i=0;i<recNum2;i++)
	{
		strcpy(curLink, rec_mk.urls[i].url);
		printf("<LI><a href=%s>%s</a>(%.2f)</LI>", curLink, curLink, rec_mk.urls[i].wt);
		if(i >=MAX_RECS_SHOW/2-1) break;
	}
	printf("<!--Markov_End-->");
	
	//modify input parameter in tandem feed collaborative filtering output to content input.
	/*for(i=0;i<rec_cf.num;i++)
	{
		strcpy(session.urls[i].url,libr_root);
		strcat(session.urls[i].url, rec_cf.urls[i].url);
	}
	if(rec_cf.num > 0)	session.num = rec_cf.num;*/
	
	//modify input parameter in tandem feed markov out to content input.
	for(i=0;i<rec_mk.num&&i<2;i++)
	{
		if(rec_mk.num < 2) strcpy(session.urls[0].url, rec_mk.urls[0].url);
		else strcpy(session.urls[1-i].url, rec_mk.urls[i].url);
	}
	if(rec_mk.num > 0)	session.num = i;
	
	//anchor text based results
	bzero(buf, sizeof(buf));
	iRet = getAnchorBasedRecs(session, buf);
	if(iRet < 0)
	{
		fprintf(stderr, "getAnchorBasedRecs failed\n");
		return -1;
	}
	if(iRet == 0)
	{
		printf("<H3>That's Strange No Anchor Text Based Recommendations</H3>");
	}
	else
	{
		printf("<H3>Anchor Text-based Recommendations</H3>");
	}
	printf("<!--Anchor_Begin-->");
	printf("%s", buf);
	printf("<!-Anchor_End-->");
		
	//content based results
	bzero(buf, sizeof(buf));
	iRet = getContentBasedRecs(session, buf);
	if(iRet < 0)
	{
		fprintf(stderr, "getContentBasedRecs failed\n");
		return -1;
	}
	printf("<H3>Content-based Recommendations</H3>");
	printf("<!--Content_Begin-->");
	printf("%s", buf);
	printf("<!--Content_End-->");
	return 0;
	
}