List <FetchedUser> FetchRemote(string vkLogin, string vkPassword)
{
var userIds = new List <long>();
foreach (var year in Years)
{
foreach (int facultyId in FacultiesExtensions.EnumerateReal())
{
var response = vkApi.Users.Search(new UserSearchParams
{
University = UniversityId,
UniversityFaculty = facultyId,
UniversityYear = year,
Count = 1000
});
userIds.AddRange(response.Select(u => u.Id));
}
}
var results = new List <FetchedUser>();
int i = 0;
foreach (var userId in userIds)
{
// any f****n vk.net error
try
{
if (UsersLimit != null && i >= UsersLimit)
{
break;
}
var infoResponse = vkApi.Call(
"execute.getUserInfo",
new VkNet.Utils.VkParameters(new Dictionary <string, string>
{
["id"] = userId.ToString()
}));
i++;
// try deserialize if wall is not hidden
try
{
var fetchedResponse = JsonConvert.DeserializeObject <FetchedResponse>(infoResponse.RawJson);
results.Add(fetchedResponse.Response);
}
catch (JsonSerializationException)
{
continue;
}
}
catch (Exception ex)
{
logger.LogWarning($"User with ID {userId} is skipped during error {ex.Message}");
continue;
}
}
return(results);
}
public Dictionary <Faculties, double> Analyze(FetchedUser user)
{
if (!fetchingTask.IsCompleted)
{
throw new NotImplementedException();
}
// form doc-term matrix
var docs = new List <Faculties>();
docs.AddRange(FacultiesExtensions.EnumerateReal());
docs.Add(Faculties.Target);
var body = users.SelectMany(u => u.GetWords())
.Where(w => w != string.Empty && w.Length > 2);
var terms = users.SelectMany(u => u.GetWords())
.Take(300)
.Distinct()
.ToArray();
var dtMat = new double[docs.Count, terms.Count()];
for (int i = 0; i < dtMat.GetLength(0); i++)
{
IEnumerable <string> facultyBody;
if (dtMat.GetLength(0) - 1 == i) // target faculty
{
facultyBody = user.GetWords();
}
else
{
facultyBody = users
.Where(u => u.Universities != null && u.Universities
.Select(uni => uni.FacultyId)
.Contains((int)docs[i]))
.SelectMany(u => u.GetWords());
}
for (int j = 0; j < dtMat.GetLength(1); j++)
{
dtMat[i, j] = facultyBody.Count(w => w == terms[j]);
}
}
svd = new SingularValueDecomposition(dtMat);
// compute low-ranked matrix
var lowRankedMat = Matrix.Dot(svd.LeftSingularVectors, svd.DiagonalMatrix); // did not low-rank here
// form index model
var indexModel = new Dictionary <Faculties, double[]>();
for (int i = 0; i < docs.Count; i++)
{
var vector = new double[svd.Rank];
for (int j = 0; j < vector.Length; j++)
{
vector[j] = lowRankedMat[i, j];
}
indexModel.Add(docs[i], vector);
}
// calculate ranges
var target = indexModel[Faculties.Target];
var facs = FacultiesExtensions.EnumerateReal();
var result = new Dictionary <Faculties, double>();
foreach (var faculty in facs)
{
var vector = indexModel[faculty];
double sum = 0;
for (int i = 0; i < svd.Rank; i++)
{
sum += Math.Pow(vector[i] - target[i], 2);
}
var range = Math.Sqrt(sum);
result.Add(faculty, range);
}
return(result.OrderBy(r => r.Value).ToDictionary(ks => ks.Key, vs => vs.Value));
}
