Source

orange-network / _network / widgets / OWPubmedView.py

Full commit
Miha Stajdohar e47d43d 


Miha Stajdohar aa20553 
Miha Stajdohar e47d43d 




































































































































































































































































"""
<name>Pubmed Network View</name>
<description></description>
<icon>icons/NetworkExplorer.svg</icon>
<contact></contact> 
<priority>6470</priority>
"""

import Orange
import OWGUI

from OWWidget import *

class PubmedNetworkView(Orange.network.NxView):
    """Network Inside View
    
    """

    def __init__(self, parent):
        Orange.network.NxView.__init__(self)

        self._nhops = 2
        self._edge_threshold = 0.5
        self._algorithm = 0 # 0 without clustering, 1 with clustering
        self._n_max_neighbors = 10
        self._center_nodes = []
        self.parent = parent
        self._hidden_nodes = []
        self._k_algorithm = 0.3
        self._delta_score = {}

    def init_network(self, graph):
        self._network = graph

        if hasattr(self.parent, 'init_network'):
            self.parent.init_network()

        if graph is None:
            return None

        return graph


    def update_network(self):
        if self._center_nodes == []:
            return

        subnet = Orange.network.Graph()
        central_nodes, to_add = self._center_nodes[:], self._center_nodes[:]
        for l in range(self._nhops):
            for i in central_nodes:
                neig = sorted([x for x in self._network.neighbors(i) if self._network.edge[i][x]['weight'] > self._edge_threshold], reverse=True)
                if len(neig) > self._n_max_neighbors:
                    neig = neig[:self._n_max_neighbors]
                to_add.extend(neig)
            central_nodes = neig
        to_add = list(set(to_add))
        subnet.add_nodes_from([(x, self._network.node[x]) for x in to_add])
        nodes = subnet.nodes()
        while nodes:
            i = nodes.pop()
            subnet.node[i] = self._network.node[i]
            neig = [x for x in self._network.neighbors(i) if x in nodes]
            subnet.add_weighted_edges_from([(i, x, w) for x, w in zip(neig, [self._network.edge[i][y]['weight'] for y in neig])])

        subnet.remove_nodes_from(self._hidden_nodes)
        subnet = self._propagate(subnet)

        if self._nx_explorer is not None:
            self._nx_explorer.change_graph(subnet)

    def set_nhops(self, nhops):
        self._nhops = nhops

    def set_edge_threshold(self, edge_threshold):
        self._edge_threshold = edge_threshold

    def set_algorithm(self, algorithm):
        self._algorithm = algorithm

    def set_n_max_neighbors(self, n_max_neighbors):
        self._n_max_neighbors = n_max_neighbors

    def set_center_nodes(self, c_nodes):
        self._center_nodes = c_nodes

    def set_k(self, k):
        self._k_algorithm = k

    def node_selection_changed(self):
        pass

    def set_user_node_score(self, node_index, new_score):
        self._delta_score[node_index] = new_score - self._network.node[node_index]['score']
        self._network.node[node_index]['score'] = new_score
        self._network.node[node_index]['user_score'] = 1

    def update_center_nodes(self, new_node):
        self._center_nodes.append(new_node)

    def _get_neighbors(self):
    #TO DELETE?
        nodes = set([self._center_node])
        for n in range(self._nhops):
            neighbors = set()
            for node in nodes:
                neighbors.update(self._network.neighbors(node))
            nodes.update(neighbors)
        return nodes

    def _propagate(self, net):
        central_nodes = [i for i in net.nodes() if net.node[i]['user_score'] == 1]
        central_delta_score = [self._delta_score[id] for id in central_nodes]
        cluster_centers = [net.node[x]['cluster'] for x in central_nodes if net.node[x]['cluster'] != '']
        updated = central_nodes[:]
        L = 0
        while True:
            to_update = []
            for i in central_nodes:
                to_update.extend(net.neighbors(i))
            to_update = set(to_update) - set(updated)
            if to_update:
                for i in to_update:
                    if not net.node[i]['user_score']:
                        predec = [j for j in net.neighbors(i) if j in central_nodes]
                        predec_delta_score = [central_delta_score[central_nodes.index(x)] for x in predec]
                        self._k_algortihm = self.compute_k(i, cluster_centers, net)
                        if self._k_algortihm: #else the node is isolated and the score doesn't change
                            net.node[i]['score'] = _prop(predec, predec_delta_score, i, L, net)
                        net.node[i]['level'] = L
                updated.extend(to_update)
                central_nodes = list(to_update)
                central_delta_score = []
                for id in central_nodes:
                    if self._delta_score.get(id, []):
                        central_delta_score.append(self._delta_score[id])
                    else:
                        central_delta_score.append(net.node[id]['score'] - 0.5)
                L += 1
            else:
                return net

    def compute_k(self, id, centers, net):
        import networkx as nx #VEDERE SE C'E' IN ORANGENET 
        if self._algorithm == 0: #no clustering
            return self._algorithm
        else:   #clustering
            paths = []
            for c in centers:
                try:
                    paths.append(nx.algorithms.shortest_paths.generic.shortest_path_length(net, id, c))
                except nx.exception.NetworkXNoPath:
                    return None #isolated node

            return float(min(paths)) / 10


    def _prop(self, predec, predec_delta_score, node, L, net):
        edges_weight = [net[i][node]['weight'] for i in predecessors]
        x = [(delta * w + 0.5) / (1 + level * self._k_algorithm) for delta, w in zip(predec_delta_score, edges_weight)]
        return max(x)


class OWPubmedView(OWWidget):

    settingsList = ['_nhops']

    def __init__(self, parent=None, signalManager=None):
        OWWidget.__init__(self, parent, signalManager, 'Pubmed Network View', wantMainArea=0)

        self.inputs = []
        self.outputs = [("Nx View", Orange.network.NxView)]

        self._nhops = 2
        self._edge_threshold = 0.5
        self._n_max_neighbors = 20
        self.selected_titles = []
        self.titles = []
        self.filter = ''
        self.ids = []
        self._selected_nodes = []
        self._algorithm = 0
        self._k_algorithm = 0.3

        self.loadSettings()

        box = OWGUI.widgetBox(self.controlArea, "Paper Selection", orientation="vertical")
        OWGUI.lineEdit(box, self, "filter", callback=self.filter_list, callbackOnType=True)
        self.list_titles = OWGUI.listBox(box, self, "selected_titles", "titles", selectionMode=QListWidget.MultiSelection, callback=self.update_view)
        OWGUI.separator(self.controlArea)
        box_pref = OWGUI.widgetBox(self.controlArea, "Preferences", orientation="vertical")
        OWGUI.spin(box_pref, self, "_nhops", 1, 6, 1, label="Number of hops: ", callback=self.update_view)
        OWGUI.spin(box_pref, self, "_n_max_neighbors", 1, 100, 1, label="Max number of neighbors: ", callback=self.update_view)
        OWGUI.doubleSpin(box_pref, self, "_edge_threshold", 0, 1, step=0.01, label="Edge threshold: ", callback=self.update_view)
        OWGUI.separator(self.controlArea)
        box_alg = OWGUI.widgetBox(self.controlArea, "Interest Propagation Algorithm", orientation="vertical")
        radio_box = OWGUI.radioButtonsInBox(box_alg, self, "_algorithm", [], callback=self.update_view)
        OWGUI.appendRadioButton(radio_box, self, "_algorithm", "Without Clustering", callback=self.update_view)
        OWGUI.doubleSpin(OWGUI.indentedBox(radio_box), self, "_k_algorithm", 0, 1, step=0.01, label="Parameter k: ", callback=self.update_view)
        OWGUI.appendRadioButton(radio_box, self, "_algorithm", "With Clustering", callback=self.update_view)

        self.inside_view = PubmedNetworkView(self)
        self.send("Nx View", self.inside_view)


    def init_network(self):
        if self.inside_view._network is None:
            return

        self.titles = [self.inside_view._network.node[node]['title'] for node in self.inside_view._network.nodes()]
        self.ids = self.inside_view._network.nodes()

        QObject.connect(self.inside_view._nx_explorer.networkCanvas, SIGNAL('point_rightclicked(Point*)'), self.node_menu_show)


    def update_view(self):
        self.inside_view.set_nhops(self._nhops)
        self.inside_view.set_edge_threshold(self._edge_threshold)
        self.inside_view.set_n_max_neighbors(self._n_max_neighbors)
        self.inside_view.set_algorithm(self._algorithm)
        self.inside_view.set_k(self._k_algorithm)
        self._selected_nodes = [self.ids[row] for row in self.selected_titles]
        self.inside_view.set_center_nodes(self._selected_nodes)
        self.inside_view.update_network()


    def filter_list(self):
        """Given a query for similar titles sets titles and ids"""

        str_input = self.filter
        str_input = str_input.strip(' .').lower().split(' ')
        titles2 = [(n, str.lower(self.inside_view._network.node[n]['title'].encode('utf-8').strip(' .')).split(' ')) for n in self.inside_view._network.nodes()] # [(id,title)]
        titles2 = sorted(titles2, key=lambda x: sum(i in str_input for i in x[1]), reverse=True)
        self.titles = [self.inside_view._network.node[x[0]]['title'] for x in titles2]
        self.ids = [x[0] for x in titles2]

    def node_menu_show(self, node):
        menu = QMenu(self)
        menu.addAction('Expand Node')
        menu.addAction('Hide Node')
        qstr = QString('Set Score')
        submenu = menu.addMenu(qstr)
        submenu.addAction('0')
        submenu.addAction('0.2')
        submenu.addAction('0.5')
        submenu.addAction('0.7')
        submenu.addAction('1')
        self.connect(menu, SIGNAL("triggered(QAction*)"), lambda action, node=node: self.node_menu_triggered(action, node))
        menu.popup(QCursor.pos())

    def node_menu_triggered(self, action, node):
        # delete_node_and_neig --> ????        
        if action.text() == 'Expand Node':
            self.inside_view.update_center_nodes(node.index())

        elif action.text() == 'Hide Node':
            self.inside_view._hidden_nodes.append(node.index())

        else:
            self.inside_view.set_user_node_score(node.index(), float(action.text()))

        self.inside_view.update_network()  #chiama propagate