Address discrepancies in docs (#36)

README.md

@@ -54,7 +54,7 @@ target = rng.rand(100,50)
 # fit and get neighbors
 k_inst = Kiez()
 k_inst.fit(source, target)
-nn_dist, nn_ind = k_inst.kneighbors()
+nn_dist, nn_ind = k_inst.kneighbors(5)
 ```
 Using (A)NN libraries and hubness reduction methods:
 ``` python
@@ -65,18 +65,16 @@ rng = np.random.RandomState(0)
 source = rng.rand(100,50)
 target = rng.rand(100,50)
 # prepare algorithm and hubness reduction
-algo_kwargs = {"n_candidates": 10}
-k_inst = Kiez(n_neighbors=5, algorithm="Faiss" algorithm_kwargs=algo_kwargs, hubness="CSLS")
+k_inst = Kiez(n_candidates=10, algorithm="Faiss", hubness="CSLS")
 # fit and get neighbors
 k_inst.fit(source, target)
-nn_dist, nn_ind = k_inst.kneighbors()
+nn_dist, nn_ind = k_inst.kneighbors(5)
 ```
 
 ## Torch Support
 Beginning with version 0.5.0 torch can be used, when using `Faiss` as NN library:
 
 ```python
-
     from kiez import Kiez
     import torch
     source = torch.randn((100,10))
@@ -89,7 +87,6 @@ Beginning with version 0.5.0 torch can be used, when using `Faiss` as NN library
 You can also utilize tensor on the GPU:
 
 ```python
-
     k_inst = Kiez(algorithm="Faiss", algorithm_kwargs={"use_gpu":True}, hubness="CSLS")
     k_inst.fit(source.cuda(), target.cuda())
     nn_dist, nn_ind = k_inst.kneighbors()

docs/index.rst

@@ -21,7 +21,7 @@ The central class of kiez serves to bundle all necessary steps to obtain nearest
     # fit and get neighbors
     k_inst = Kiez()
     k_inst.fit(source, target)
-    nn_dist, nn_ind = k_inst.kneighbors()
+    nn_dist, nn_ind = k_inst.kneighbors(5)
 
 The main feature of kiez lies in the ability to use hubness reduction methods and approximate nearest neighbor (ANN) algorithms. This enables you to profit from the speed advantage of ANN algorithms, while achieving highly accurate nearest neighbor results:
 
@@ -34,14 +34,10 @@ The main feature of kiez lies in the ability to use hubness reduction methods an
     source = rng.rand(100,50)
     target = rng.rand(100,50)
     # prepare algorithm and hubness reduction
-    from kiez.neighbors import HNSW
-    hnsw = HNSW(n_candidates=10)
-    from kiez.hubness_reduction import CSLS
-    hr = CSLS()
+    k_inst = Kiez(n_candidates=10, algorithm="Faiss", hubness="CSLS")
     # fit and get neighbors
-    k_inst = Kiez(n_neighbors=5, algorithm=hnsw, hubness=hr)
     k_inst.fit(source, target)
-    nn_dist, nn_ind = k_inst.kneighbors()
+    nn_dist, nn_ind = k_inst.kneighbors(5)
 
 You can install kiez via pip:
 

docs/source/usage.rst

@@ -9,8 +9,8 @@ The `Kiez` class enables the usage of different nearest neighbor (NN) algorithms
 
     # via string and arguments as dict
     k_inst = Kiez(
-        algorithm="HNSW",
-        algorithm_kwargs={"n_candidates": 10},
+        algorithm="SklearnNN",
+        n_candidates=10,
         hubness="LocalScaling",
         hubness_kwargs={"method": "NICDM"},
     )
@@ -20,8 +20,8 @@ The `Kiez` class enables the usage of different nearest neighbor (NN) algorithms
     from kiez.neighbors import HNSW
 
     k_inst = Kiez(
-        algorithm=HNSW,
-        algorithm_kwargs={"n_candidates": 10},
+        algorithm=SklearnNN,
+        n_candidates=10,
         hubness=LocalScaling,
         hubness_kwargs={"method": "NICDM"},
     )
@@ -35,7 +35,7 @@ The `Kiez` class enables the usage of different nearest neighbor (NN) algorithms
 
     # content of 'conf.json' file
     # {
-    #   "algorithm": "HNSW",
+    #   "algorithm": "SklearnNN",
     #   "algorithm_kwargs": {
     #     "n_candidates": 10
     #   },
@@ -57,17 +57,10 @@ With your initialized kiez instance you are ready to fit your data and retrieve
     source = rng.rand(100,50)
     target = rng.rand(100,50)
     k_inst.fit(source, target)
-    neigh_dist, neigh_ind = k_inst.kneighbors()
+    neigh_dist, neigh_ind = k_inst.kneighbors(5)
 
 This will retrieve all nearest neighbors of source entities in the target entities.
 
-You can also query for specific entities and a specific number of k neighbors:
-
-.. code-block:: python
-
-    neigh_dist, neigh_ind = k_inst.kneighbors()
-    # get 2 nearest neighbors of the first 5 source entities
-    k_inst.kneighbors(source_query_points=source[:5,:], k=2)
 
 Single source case
 -------------------
@@ -82,8 +75,6 @@ While the main focus of kiez is to be part of an embedding-based entity resoluti
 
     # get the nearest neighbors of all source entities amongst themselves
     k_inst.kneighbors()
-    # get 2 nearest neighbors of the first 5 source entities
-    k_inst.kneighbors(source_query_points=source[:5,:], k=2)
 
 Evaluation
 ----------

kiez/kiez.py

@@ -23,8 +23,8 @@ class Kiez:
 
     Parameters
     ----------
-    n_neighbors : int, default=5
-        number of nearest neighbors used in search
+    n_candidates : int, default=10
+        number of nearest neighbors used for candidate search
     algorithm : :obj:`~kiez.neighbors.NNAlgorithm`, default = None
         initialised `NNAlgorithm` object that will be used for neighbor search
         If no algorithm is provided :obj:`~kiez.neighbors.Faiss` is used if available else
@@ -53,7 +53,7 @@ class Kiez:
     >>> # fit and get neighbors
     >>> k_inst = Kiez()
     >>> k_inst.fit(source, target)
-    >>> nn_dist, nn_ind = k_inst.kneighbors()
+    >>> nn_dist, nn_ind = k_inst.kneighbors(5)
 
     Using a specific algorithm and hubness reduction
 
@@ -64,18 +64,10 @@ class Kiez:
     >>> source = rng.rand(100,50)
     >>> target = rng.rand(100,50)
     >>> # prepare algorithm and hubness reduction
-    >>> from kiez.neighbors import NMSLIB
-    >>> hnsw = NMSLIB(n_candidates=10)
-    >>> from kiez.hubness_reduction import CSLS
-    >>> hr = CSLS()
+    >>> k_inst = Kiez(n_candidates=10, algorithm="Faiss", hubness="CSLS")
     >>> # fit and get neighbors
-    >>> k_inst = Kiez(n_neighbors=5, algorithm=hnsw, hubness=hr)
     >>> k_inst.fit(source, target)
-    >>> nn_dist, nn_ind = k_inst.kneighbors()
-
-    NN and hubness algorithms can also be supplied via string:
-
-    >>> k_inst = Kiez(algorithm="SklearnNN", hubness="CSLS")
+    >>> nn_dist, nn_ind = k_inst.kneighbors(5)
 
     You can investigate which NN algos are installed and which hubness methods are implemented with:
 

tests/example_conf.json

@@ -1,5 +1,5 @@
 {
-  "algorithm": "NMSLIB",
+  "algorithm": "SklearnNN",
   "algorithm_kwargs": {
     "n_candidates": 10
   },

tests/test_kiez.py

@@ -121,7 +121,9 @@ def test_from_config():
         assert isinstance(kiez.hubness, LocalScaling), f"wrong hubness: {kiez.hubness}"
         assert kiez.algorithm is not None
         assert isinstance(kiez.algorithm, NNAlgorithm)
-        assert isinstance(kiez.algorithm, NMSLIB), f"wrong algorithm: {kiez.algorithm}"
+        assert isinstance(
+            kiez.algorithm, SklearnNN
+        ), f"wrong algorithm: {kiez.algorithm}"
 
 
 def mock_make(name, algorithm_kwargs):