Inferring borrower network in a microfinancing framework (KIVA)
Citations
3 citations
Cites background or methods from "Inferring borrower network in a mic..."
...One of the solutions suggested and implemented by earlier researches is the usage of digital technology to support and enhance MFI services (Augburg, Schmidt, and Krishnaswainy, 2011; Bada, 2012; Ghosh and Vachery, 2016; Mbogo, 2010: Paruthi, Frias-Martinez, and Frias-Martinez, 2016; Sathe and Desai, 2006; Weber, Kulkarni, and Riggins, 2012)....
[...]
...Consequently, digital technology has been used to enhance the services provided by MFIs, such as KIVA, the world’s first online lending platform connecting online lenders to entrepreneurs across the globe (Ghosh and Vachery, 2016; Paruthi et al., 2016; Weber et al., 2012)....
[...]
...Nature of the Interaction between Microfinance Institutions and Micro Businesses The interaction between MBs and MFIs is based mostly on relationship lending, which replaces the collateral need for low-income earners to qualify for loans (Ghosh and Vachery, 2016; Uchida et al., 2012)....
[...]
...The interaction between MBs and MFIs is based mostly on relationship lending, which replaces the collateral need for low-income earners to qualify for loans (Ghosh and Vachery, 2016; Uchida et al., 2012)....
[...]
...Specifically, mobile technology can reduce the number of middlemen, as in peer-to-peer lending and KIVA, which eliminate intermediaries between MFIs and borrowers, and present a viable solution to the agency problem (Ghosh and Vachery, 2016; Paruthi et al., 2016)....
[...]
References
16,176 citations
"Inferring borrower network in a mic..." refers methods in this paper
...We use the KullbackLeibler divergence (KL) Divergence metric (A measure of dissimilarity between two probability distributions) based on [5] to estimate the similarity between top-k pairs of lenders like number of loans lent on other attributes like distribution among sectors lent (Sector KL Divergence), distribution over type of loans lent i....
[...]
2,036 citations
"Inferring borrower network in a mic..." refers methods in this paper
...The time complexity for each iteration of SimRank is O(n2d) where d is the average of |I(A)||I(B)| or |O(A)||O(B)| or both as applicable over all nodepairs (A,B)....
[...]
...We formulate a novel tripartite extension of SimRank using the network of lenders, loans and borrowers to capture the inherent pattern in the system....
[...]
...We can extend this formulation to a general Multipartite SimRank....
[...]
...In addition, we also propose a Multipartite extension to SimRank motivated by real life applications....
[...]
...Since we see that the tripartite SimRank formulation works well in modeling similarity of lenders and loans, it would be justifiable to assume that the disambiguation scheme adopted for borrowers is suitable which is further strengthened by 2016 IEEE International Conference on Data Science and Engineering (ICDSE) our results on borrowers....
[...]
76 citations
"Inferring borrower network in a mic..." refers background in this paper
...I. INTRODUCTION The concept of Microfinancing was first introduced by Muhammad Yunus[1], when he found that the extremely poor barely had enough means to sustain themselves....
[...]
...The concept of Microfinancing was first introduced by Muhammad Yunus[1], when he found that the extremely poor barely had enough means to sustain themselves....
[...]
44 citations
"Inferring borrower network in a mic..." refers background in this paper
..., 2014 [3], the authors propose a fairness-aware recommendation system based on one-class collaborativefiltering techniques (since only partial binary decision labels are available) where the goal in addition to maximizing the opportunity of successful matching is also diversifying the resources of loan providers....
[...]
...As noted in [3], lenders have very strong individual preferences over lenders and loans and the distribution is in no way uniform....
[...]
...In Lee et al., 2014 [3], the authors propose a fairness-aware recommendation system based on one-class collaborativefiltering techniques (since only partial binary decision labels are available) where the goal in addition to maximizing the opportunity of successful matching is also diversifying the…...
[...]