public allResults gatherAllData(string inputkeyword)
{
allResults allData = new allResults();
float competition = -1;
try
{
//fetch results page for desired keyword ONE time to prevent bans
//Note: SEOTC is a completely separate search (intitle: keyword)
string resultsPage = getSERP(inputkeyword);
//get top 10 urls
List <string> top10 = getTop10(resultsPage);
//this will store the urls plus the extra information we need
List <topSite> top10Details = new List <topSite>();
foreach (string site in top10)
{
topSite newSite = new topSite();
newSite.setPR(fetchPR(site));
newSite.setAge(fetchDA(site));
newSite.setURL(site);
//newSite.setBacklinks =
newSite.setDmozListed(checkDMOZ(site));
top10Details.Add(newSite);
}
//analyze keyword counts for every site in top 10
top10Details = analyzeKeywordCount(top10Details, inputkeyword);
keyword keyword = new keyword();
keyword.setName(inputkeyword);
keyword.setSeoc(fetchSEOC(keyword.getName()));
keyword.setSeotc(fetchSEOTC(keyword.getName()));
allData.setKeyword(keyword);
allData.setTop10(top10Details);
//competition = analyzeComp(top10Details, inputkeyword);
}
catch (Exception ex)
{
}
return(allData);
}
public void setKeyword(keyword keyword)
{
keyword_ = keyword;
}
public allResults gatherAllData(string inputkeyword)
{
allResults allData = new allResults();
float competition = -1;
try
{
//fetch results page for desired keyword ONE time to prevent bans
//Note: SEOTC is a completely separate search (intitle: keyword)
string resultsPage = getSERP(inputkeyword);
//get top 10 urls
List<string> top10 = getTop10(resultsPage);
//this will store the urls plus the extra information we need
List<topSite> top10Details = new List<topSite>();
foreach (string site in top10)
{
topSite newSite = new topSite();
newSite.setPR(fetchPR(site));
newSite.setAge(fetchDA(site));
newSite.setURL(site);
//newSite.setBacklinks =
newSite.setDmozListed(checkDMOZ(site));
top10Details.Add(newSite);
}
//analyze keyword counts for every site in top 10
top10Details = analyzeKeywordCount(top10Details, inputkeyword);
keyword keyword = new keyword();
keyword.setName(inputkeyword);
keyword.setSeoc(fetchSEOC(keyword.getName()));
keyword.setSeotc(fetchSEOTC(keyword.getName()));
allData.setKeyword(keyword);
allData.setTop10(top10Details);
//competition = analyzeComp(top10Details, inputkeyword);
}
catch (Exception ex)
{
}
return allData;
}
//***analyzeComp() details***
//===========================
//Here we will determine an estimation of how difficult it will be to rank for a certain keyword.
//For now we will keep it simple and only analyze a few metrics:
//Domain Age - How long the domain been registered for
//Page Rank (PR) - Google's page rank for the top 10 site
//SEO Comp (SEOC) - The total number of webpages GLOBALLY that mention the same keyword term
// in the same (phrase) word order, in Bing's index. [AS DESCRIBED IN MARKET SAMURAI]
//Title Comp (SEOTC) - The total number of webpages GLOBALLY that mention all of the words in a keyword
// term in the title of a page. [AS DESCRIBED IN MARKET SAMURAI]
//Possible other metrics to determine rank to be implemented:
//Search for sites such as blogger, yahoo answers or similar sites that are easy to beat
//Search for certain top level domains (TLDs) which are difficult to beat such as .gov and .edu
//With any extra time, we will add more metrics to take a look at, but for now, this should be enough to get
//a general overview of how difficult it might be to rank for the desired keyword.
//We will be using a simple scale between 1 and 10 to determine difficulty. Higher domain age/competition will add points,
//meaning a higher difficulty.
//We will get a score between 1 to 10 for each metric and then get an average between them.
//The ranks will be as follows:
//1 - Incredibly easy - Probably instant rankings
//2 - Easy - Little work necessary to rank
//3-5 - Medium - You should have some SEO knowledge to rank for these keywords
//6-8 - Hard - Probably a long time endeavour to rank for this keyword. Major SEO/time needed
//9-10 - Impossible - Probably not worth your time
//===================================================================================
// ***Update notes as of Monday, Oct 29th 2012(Sprint 2):***
//===================================================================================
//A new ranking system will be needed as the current one does not work as intended.
//In this new system, each metric will have an associated "weight".
//This is because some metrics are more important than others.
public float analyzeComp(List<topSite> top10, keyword keyword)
{
keywordMetrics newKeywordMetrics = new keywordMetrics();
float compScore = -1;
try
{
//get average pr/domain age from top10
int avgPR = 0;
int avgDA = 0;
foreach (topSite site in top10)
{
if (site.getPR() > 0) //dont add negative prs cause by errors
avgPR += site.getPR();
avgDA += site.getAge();
//SPRINT 3====================================
//more important metrics are tallied differently
if (site.getDmozListed() == true)
newKeywordMetrics.setDmozListedScore(newKeywordMetrics.getDmozListedScore() + 1);
if (site.getIsEMD() == true)
newKeywordMetrics.setIsEMDScoreScore(newKeywordMetrics.getIsEMDScoreScore() + 3); //3 for emd
if (site.getKeywordInMeta() == true)
newKeywordMetrics.setKeywordInMetaScore(newKeywordMetrics.getKeywordInMetaScore() + 1);
if (site.getKeywordInTitle() == true)
newKeywordMetrics.setKeywordInTitleScore(newKeywordMetrics.getKeywordInTitleScore() + 3); //3 for title
if (site.getKeywordInURL() == true)
newKeywordMetrics.setKeywordInURLScore(newKeywordMetrics.getKeywordInURLScore() + 1);
//END SPRINT 3================================
}
avgPR = avgPR / top10.Count;
avgDA = avgDA / top10.Count;
//pr comp
newKeywordMetrics.setPrScore(avgPR);
//da comp
if (avgDA >= 10)
newKeywordMetrics.setDaScore(10);
else
newKeywordMetrics.setDaScore(avgDA);
//seoc comp
if (keyword.getSeoc() >= 1000000)
newKeywordMetrics.setSeocScore(10);
else if (keyword.getSeoc() >= 500000)
newKeywordMetrics.setSeocScore(8);
else if (keyword.getSeoc() >= 250000)
newKeywordMetrics.setSeocScore(6);
else if (keyword.getSeoc() >= 100000)
newKeywordMetrics.setSeocScore(4);
else if (keyword.getSeoc() >= 60000)
newKeywordMetrics.setSeocScore(2);
else if (keyword.getSeoc() >= 30000)
newKeywordMetrics.setSeocScore(1);
else
newKeywordMetrics.setSeocScore(0);
//seotc comp
if (keyword.getSeotc() >= 100000)
newKeywordMetrics.setSeotcScore(10);
else if (keyword.getSeotc() >= 50000)
newKeywordMetrics.setSeotcScore(8);
else if (keyword.getSeotc() >= 25000)
newKeywordMetrics.setSeotcScore(6);
else if (keyword.getSeotc() >= 10000)
newKeywordMetrics.setSeotcScore(4);
else if (keyword.getSeotc() >= 6000)
newKeywordMetrics.setSeotcScore(2);
else if (keyword.getSeotc() >= 3000)
newKeywordMetrics.setSeotcScore(1);
else
newKeywordMetrics.setSeotcScore(0);
//max percentage for each metric variables
//SPRINT 3 NOTE
//Because of the way competition can be judged, the total percentages will exceed 100
//This is because if a certain metrics FAR outweighs all the rest, the site could be ranking
//based on that metric alone.
float maxPrScore = 5;
float maxDaScore = 15;
float maxSeocScore = 10;
float maxSeotcScore = 20;
//sprint 3 percentages
float maxDmozScore = 10;
float maxEMDScore = 10;
float maxMetaScore = 5;
float maxTitleScore = 20;
float maxURLScore = 5;
//get competition
compScore = newKeywordMetrics.getPrScore() * (maxPrScore / 100) +
newKeywordMetrics.getDaScore() * (maxDaScore / 100) +
newKeywordMetrics.getSeocScore() * (maxSeocScore / 100) +
newKeywordMetrics.getSeotcScore() * (maxSeotcScore / 100) +
newKeywordMetrics.getDmozListedScore() * (maxDmozScore / 100) +
newKeywordMetrics.getIsEMDScoreScore() * (maxEMDScore / 100) +
newKeywordMetrics.getKeywordInMetaScore() * (maxMetaScore / 100) +
newKeywordMetrics.getKeywordInTitleScore() * (maxTitleScore / 100) +
newKeywordMetrics.getKeywordInURLScore() * (maxURLScore / 100);
}
catch (Exception ex)
{
compScore = -1;
}
return compScore;
}
public void setKeyword(keyword keyword)
{
keyword_ = keyword;
}
}//end getTop10
//***analyzeComp() details***
//===========================
//Here we will determine an estimation of how difficult it will be to rank for a certain keyword.
//For now we will keep it simple and only analyze a few metrics:
//Domain Age - How long the domain been registered for
//Page Rank (PR) - Google's page rank for the top 10 site
//SEO Comp (SEOC) - The total number of webpages GLOBALLY that mention the same keyword term
// in the same (phrase) word order, in Bing's index. [AS DESCRIBED IN MARKET SAMURAI]
//Title Comp (SEOTC) - The total number of webpages GLOBALLY that mention all of the words in a keyword
// term in the title of a page. [AS DESCRIBED IN MARKET SAMURAI]
//Possible other metrics to determine rank to be implemented:
//Search for sites such as blogger, yahoo answers or similar sites that are easy to beat
//Search for certain top level domains (TLDs) which are difficult to beat such as .gov and .edu
//With any extra time, we will add more metrics to take a look at, but for now, this should be enough to get
//a general overview of how difficult it might be to rank for the desired keyword.
//We will be using a simple scale between 1 and 10 to determine difficulty. Higher domain age/competition will add points,
//meaning a higher difficulty.
//We will get a score between 1 to 10 for each metric and then get an average between them.
//The ranks will be as follows:
//1 - Incredibly easy - Probably instant rankings
//2 - Easy - Little work necessary to rank
//3-5 - Medium - You should have some SEO knowledge to rank for these keywords
//6-8 - Hard - Probably a long time endeavour to rank for this keyword. Major SEO/time needed
//9-10 - Impossible - Probably not worth your time
//===================================================================================
// ***Update notes as of Monday, Oct 29th 2012(Sprint 2):***
//===================================================================================
//A new ranking system will be needed as the current one does not work as intended.
//In this new system, each metric will have an associated "weight".
//This is because some metrics are more important than others.
public float analyzeComp(List <topSite> top10, keyword keyword)
{
keywordMetrics newKeywordMetrics = new keywordMetrics();
float compScore = -1;
try
{
//get average pr/domain age from top10
int avgPR = 0;
int avgDA = 0;
foreach (topSite site in top10)
{
if (site.getPR() > 0) //dont add negative prs cause by errors
{
avgPR += site.getPR();
}
avgDA += site.getAge();
//SPRINT 3====================================
//more important metrics are tallied differently
if (site.getDmozListed() == true)
{
newKeywordMetrics.setDmozListedScore(newKeywordMetrics.getDmozListedScore() + 1);
}
if (site.getIsEMD() == true)
{
newKeywordMetrics.setIsEMDScoreScore(newKeywordMetrics.getIsEMDScoreScore() + 3); //3 for emd
}
if (site.getKeywordInMeta() == true)
{
newKeywordMetrics.setKeywordInMetaScore(newKeywordMetrics.getKeywordInMetaScore() + 1);
}
if (site.getKeywordInTitle() == true)
{
newKeywordMetrics.setKeywordInTitleScore(newKeywordMetrics.getKeywordInTitleScore() + 3); //3 for title
}
if (site.getKeywordInURL() == true)
{
newKeywordMetrics.setKeywordInURLScore(newKeywordMetrics.getKeywordInURLScore() + 1);
}
//END SPRINT 3================================
}
avgPR = avgPR / top10.Count;
avgDA = avgDA / top10.Count;
//pr comp
newKeywordMetrics.setPrScore(avgPR);
//da comp
if (avgDA >= 10)
{
newKeywordMetrics.setDaScore(10);
}
else
{
newKeywordMetrics.setDaScore(avgDA);
}
//seoc comp
if (keyword.getSeoc() >= 1000000)
{
newKeywordMetrics.setSeocScore(10);
}
else if (keyword.getSeoc() >= 500000)
{
newKeywordMetrics.setSeocScore(8);
}
else if (keyword.getSeoc() >= 250000)
{
newKeywordMetrics.setSeocScore(6);
}
else if (keyword.getSeoc() >= 100000)
{
newKeywordMetrics.setSeocScore(4);
}
else if (keyword.getSeoc() >= 60000)
{
newKeywordMetrics.setSeocScore(2);
}
else if (keyword.getSeoc() >= 30000)
{
newKeywordMetrics.setSeocScore(1);
}
else
{
newKeywordMetrics.setSeocScore(0);
}
//seotc comp
if (keyword.getSeotc() >= 100000)
{
newKeywordMetrics.setSeotcScore(10);
}
else if (keyword.getSeotc() >= 50000)
{
newKeywordMetrics.setSeotcScore(8);
}
else if (keyword.getSeotc() >= 25000)
{
newKeywordMetrics.setSeotcScore(6);
}
else if (keyword.getSeotc() >= 10000)
{
newKeywordMetrics.setSeotcScore(4);
}
else if (keyword.getSeotc() >= 6000)
{
newKeywordMetrics.setSeotcScore(2);
}
else if (keyword.getSeotc() >= 3000)
{
newKeywordMetrics.setSeotcScore(1);
}
else
{
newKeywordMetrics.setSeotcScore(0);
}
//max percentage for each metric variables
//SPRINT 3 NOTE
//Because of the way competition can be judged, the total percentages will exceed 100
//This is because if a certain metrics FAR outweighs all the rest, the site could be ranking
//based on that metric alone.
float maxPrScore = 5;
float maxDaScore = 15;
float maxSeocScore = 10;
float maxSeotcScore = 20;
//sprint 3 percentages
float maxDmozScore = 10;
float maxEMDScore = 10;
float maxMetaScore = 5;
float maxTitleScore = 20;
float maxURLScore = 5;
//get competition
compScore = newKeywordMetrics.getPrScore() * (maxPrScore / 100) +
newKeywordMetrics.getDaScore() * (maxDaScore / 100) +
newKeywordMetrics.getSeocScore() * (maxSeocScore / 100) +
newKeywordMetrics.getSeotcScore() * (maxSeotcScore / 100) +
newKeywordMetrics.getDmozListedScore() * (maxDmozScore / 100) +
newKeywordMetrics.getIsEMDScoreScore() * (maxEMDScore / 100) +
newKeywordMetrics.getKeywordInMetaScore() * (maxMetaScore / 100) +
newKeywordMetrics.getKeywordInTitleScore() * (maxTitleScore / 100) +
newKeywordMetrics.getKeywordInURLScore() * (maxURLScore / 100);
}
catch (Exception ex)
{
compScore = -1;
}
return(compScore);
}