Socialiser - connect people in companies

A submission by team "BembelEngineers" to START Hack 2024 - CISCO Challenge

Team "BembelEngineers" from TU Darmstadt and ETH Zürich. Members: Kevin Riehl, Leon Bernard, Benedikt Völker

We present "Socialiser", a recommender system that brings people together in companies. Based on WiFi-Localisation data, we learn a social network graph, that is analysed to identify gaps in corporate network structure and actively suggests informal meetings for exchange and getting together.

How it works

• Event Stream Data from CISCO Spaces is produced.
• This Stream Data is transmitted via Firehose API to our first software module "Graph Generator".
• The "Graph Generator" translates the event stream data into a social network graph (either real time or ex-post).
• The social graph yields insights for management to define company goals.
• The company goals together with the social graph are the input for the software module "Recommender System".
• The "Recommender Software" generates meeting suggestions to actively connect people to achieve company goals.
• The meeting suggestions are then sent via the WebEx Messaging Bot API to chats of the peopl in the company.
• These meeting suggestions lead to real meetings and thus serve the company goals.

The calculated centralities and a possible recommendation:

katz_centrality = [0.29344988 0.29344988 0.32605542 0.2974038  0.2974038  0.33809951 0.34483903 0.32605542 0.29344988 0.34483903]
degree_centrality = [0.         0.         0.66538462 0.08846154 0.08846154 0.86153846 1.         0.66538462 0.         1.        ]

influential_node = 7
isolated_node = 1
recommended_meeting = [influential_node, isolated_node]

Structure of this repository

In this repository you will find three software modules.

• Folder graphGenerator contains a Python project that processes StreamData from CISCO Spaces via Firehose API, and generates the social network graph.
• Folder recommenderSystem contains a Python project that analyses the social network graph and generates meeting recommendations.
• Folder webexAPI contains a NodeJS project that enables to send meeting invites automatically via a WebexChat Bot.

Exemplary Code Workflow

# Record Stream Data
apiKey = establishConnection()
fileWriter = startLogging(targetFile="logs.json")
recordStream(apiKey, fileWriter)

# Filter Stream Data
fullStream = loadData(targetFile="logs.json")
filteredStream = filterStreamData(fullStream)

# Convert To User-Timeline
timelineDF = generateTimeLine(filteredStream2)

# Calculate Social Network Graph
graph_matrix = generateGraph(timelineDF)

# Generate a undirected graph with weighted edges
graph = create_graph(graph_matrix)

# Calculate the katz centrality for the graph nodes
katz = centrality_katz(graph)

# Calculate the degree centrality for the graph nodes
degr = centrality_degree(graph)

# Evalutate the most influencial and most isolated persons
infl, isol = recommend_meeting(katz, degr)