def digest(self,
search_result: [str],
click: {str: int},
norm=10.0) -> np.array:
if self.dim_search is not None:
vec1 = [0.0] * self.dim_search
for path in search_result:
if '/' not in path:
return None
cat_ids = path.split('/')[1:]
for i, d in enumerate(cat_ids):
lv = i + 1
index = self.catid2index[d] % self.dim_search
value = 1.0 if farmhash.hash64withseed(
d, 4321) % 2 == 0 else -1.0
vec1[index] += value / lv
vec2 = [0.0] * self.dim_clicks
for path, count in click.items():
if '/' not in path:
return None
cat_ids = path.split('/')[1:]
for i, d in enumerate(cat_ids):
lv = i + 1
index = self.catid2index[d] % self.dim_clicks
value = 1.0 if farmhash.hash64withseed(d,
4321) % 2 == 0 else -1.0
vec2[index] += value / lv * count
if self.dim_search is not None:
vec1 = np.array(vec1)
vec2 = np.array(vec2)
ret = np.concatenate((vec1, vec2))
else:
ret = vec2
ret = ret / (np.linalg.norm(ret) / norm)
if np.isnan(ret).any():
return None
return ret
def get_signatures(self, queries) -> [[str]]:
signatures = [] # size == self.L (table count)
table_values = self.digest_all_tables(queries)
# table_values: signaures across L hash tables
for table_value in table_values:
table_value = map(
lambda bits: ''.join(map(str, bits)), table_value)
if self.bucket_limit is not None:
table_value = map(
lambda v: str(
farmhash.hash64withseed(v, 2048) % self.bucket_limit),
table_value
)
table_value = list(table_value)
signatures.append(table_value)
return signatures
def get_signatures(self, queries) -> [[str]]:
# length==self.L*self.F
# shape==(self.L*self.F) * N_queries
signatures = []
# table_values: signaures across L hash tables
table_values = self.digest_all_tables(queries)
# table_value: an (F*N_queries) * K matrix
for table_value in table_values:
table_value = map(
lambda bits: ''.join(map(str, bits)), table_value)
if self.bucket_limit is not None:
table_value = map(
lambda v: str(
farmhash.hash64withseed(v, 2048) % self.bucket_limit),
table_value
)
# length==(F*N_queries)
table_value = np.array(list(table_value))
# shape== F * N_queries
table_value = table_value.reshape(-1, self.F).T
signatures.extend(table_value)
assert len(signatures[-1]) == len(queries), "invalid shape on sigs axis 1"
return signatures
def __call__(self, x):
"""Returns the hash x modulus self.modulus."""
val = farmhash.hash64withseed(str(x), self._random_seed)
return val if self.modulus is None else val % self._modulus