Mentions Filtering

user666be710f417d

A new feature needed is to make the existing system work better for me a user. I have received numerous references with the question " Is this paper co-authored by you". I have each time responded yes or no but I keep getting the same email about the same papers as if my reply makes no difference...

5 months ago (edited)

1

reply

hannah_ehrlich

Merged in a post:

1

Need feature to improve accuracy of mentions

The mentions are a total disaster. I don't know if you can use AI or if you can improve your algorithm in some way, but after I tell Academia ten times that I don't publish on tracheotomies or well-being of first-gen college students, it should learn to stop proposing papers to me as "mentions" that involve tracheotomies and/or the well-being of first-gen college students. (This is partly an extesion of the problem caused by your extremely stupid idea of using people's first initials and last names as if they indicated actual mentions. But how many people named "A. Smith" are there in the entire world? And do you really, truly think that every single A. Smith is the same person? Who has published a million papers on everything from tracheotomies to trans studies, enzyme breakdown to Turkish literature, byproducts of petroleum refining to avant-garde Chinese cinema?)

user669dcedeba0bc 3 months ago New Features

4 months ago

0

hannah_ehrlich

Merged in a post:

4 months ago

0

hannah_ehrlich

Merged in a post:

1

Incorrect Notifications

Mentions/Notifications You send me emails saying my name has been mentioned. 55 of them. They are all incorrect. example: Hong S-B, Hubka V, Klaassen CHW, This is Hubka V. But, you take the B from the previous authorr and say it is me. This is very aggravating. Please fix your system. Brian Hubka

bghubka 4 months ago New Features

4 months ago

0

hannah_ehrlich

Merged in a post:

1

Never-ending Mention problems

I have written this many times over the years. I have not received a response. I even wrote to the personal accounts of the administrators on Linkedin. I still have not received a response. Academia is going back a little more every day. Academia engineers insist on living in 1985 even though we are in the year 2024. I am tired of saying this. However, I will write it for the last time. I was a high school student in 2004. However, Academia sends me an article written in 2004 and asks "is this you?" I am a historian. However, Academia shows me articles on religion, nuclear missiles, literature, foreign trade, social media and asks "is this you?" Why? Why? Why? The number of my approved mentions is 84. The number of my rejected mentions is 3,554. Have you never thought about this failure? Have you never tried to solve this stupid problem? I have written many times. All academics are fed up with you. All academics use academia out of necessity and they hate you. The year doesn't match. My career doesn't match. The subject doesn't match. The field of study doesn't match. Sometimes even the name doesn't match. Not everything can be wrong. But in academia, it's possible.

alibalci 3 months ago New Features

4 months ago

0

hannah_ehrlich

changed category from "New Features" to "Product Improvements"

4 months ago

1

hannah_ehrlich

Merged in a post:

1

"Shanti.S" is not "Shiva Shakti Shanti" *

Hello, i'm french and write in french, i receive dozens of notifications telling me that "Shanti.S" was mentioned in english, whereas I'm "Shiva Shakti Shanti" in french, i know i've got a simple non-premium account but it's a pain, and it means a lot about your algorithme neo-colonialist = would it tag any "George" at any time it is mentioned as first name in an academia's article, as a last name ? I don't reckon, no. So be aware about the fact "shanti" is one of the most common first name in south-india, you've got one shanti per family, as in your "christian" western countries you've got as many "marie" . Please make learn your algorithme, other than in the old patterns ways ; ; ; shanti (shiva shakti shanti - france)

shanti999 3 months ago Product Improvements

4 months ago

0

aliciaanderson

Citation styles only vary so much. An APA citation or an MLA or Chicago citation for a single paper should only generate maybe 1-3 regular expressions that tie a paper to all citations that reference it. Our mentions should be disavowed or not by the publication, not by individual citations. Instead of thousands of too-broad keyword searches for our names, we should be able to say "I did not write this paper" once, and never have it or any citations referring to it appear back in our feeds.

3 months ago (edited)

0

reply

hannah_ehrlich

Merged in a post:

1

Are you the author? is almost worse than useless

This is my personal experience but I'm sure it will apply to the majority of users. I regularly get "Are you the author?" emails but as there are more than 1000 matches, it's not practical for me to go through and inform academia.edu that none of them are my references. My academic name is James F. R. Masters and I use this specifically to distinguish me from other authors with firstname James and surname Masters. Unfortunately, academia.edu matches me against anyone with firstname James and surname Masters, rendering the matching algorithm worse than useless (as it wastes my time and produces only spurious statistics and notifications). There is an obvious and easy improvement that can be made here - allow the users to choose what strings to match against. If I could simply configure the matching algorithm to only match against 'James F. R. Masters' or 'Masters, James F. R.' or 'Masters, J. F. R', say, then everything would work as intended and I would be able to get an actual valid and useful result as to how many citations I have.

jfrm 2 months ago Product Improvements

2 months ago

0

peter

I agree that the ‘Mentions’ function is not fit for purpose because it is too loose. I have now unsubscribed from it as I am bombarded with hundreds of references to people with names similar to mine.

I have just received an email from Academia.edu asking me to confirm the set of permutations of my name or to add more permutations. There was no option to disconfirm any of the permutations, which was the only option I wished to exercise. I have to assume that this questionnaire was driven by some manager who thinks the objective is to maximise the number of theoretically possible references. So the email proudly tells me that 1,400+ mentions have been found. No! Almost all of these are not me, and most of them could be eliminated by a slightly more intelligent software.

My specific suggestions follow.

Allow the user to disconfirm permutations of name. In my own case, I always sign my name in publications as “Peter B Lloyd”. Depending of journal styles, this will be cited either as “Peter B Lloyd” or “P B Lloyd”. Any reference to “Peter Lloyd” or “P Lloyd” is guaranteed not to be me.
Take account of hyphenated names. There is a very prolific author called “Peter Lloyd-Sherlock” and I get all of his citations. (And I assume he gets mine.) a simple change to the software would fix this.
Allow the user to white-list or black-list subject areas. For example any publications by “P(eter) B Lloyd” in biochemistry or housing is not me.

If the referenced publication could be traced by the software then further filtering options arise:

Filter by ORCID. If an author has a different ORCID from mine, then s/he ain’t me.
Filter by institution. A slightly more intelligent software coukd determine the author’s institution in the referenced publication, and the user can exclude a lot of false matches.

a month ago (edited)

0

reply

hannah_ehrlich

Merged in a post:

1

More on Mentions-filtering

I get literally thousands of suggested mentions -- far too many to go though -- but most of them are inapproporiate and the system is crying out for some coarse-grained filltering to make the effort practical. For a start, the majority of the suggested mentions that I get are from medical or geological papers, and I would like to eliminate them entirely in one click. All possible mentions should have a "field" property which can be used as a filter. Similarly with papers or books published before I was born, or before my own publications. Lastly, the name matching is woeful and basically includes anything with my surname and at least one initial, somewhere near it. It seems incapablle of distinguishing schemes where my forenames follow my surname, after a comma, or precede it. I use a consistent format for my formal name and would suggest at least a way of weighting the variations on my account such that I can distinguish clear cases of me from possible ones, and from extremely unlikely ones.

parallaxview 1 month ago Product Improvements

a month ago

0

hannah_ehrlich

Merged in a post:

1

names are names

My name is Bruno VAN de VLIET, not Bruno, not Bruno Van, not Bruno Van de, only Bruno VAN de VLIET. I publish under my name and sometimes I add Dr. Why do I get socalled mentions with other name or part of my name. This is disturbing and it is wrong. I tried to opt out of it but your system does not allow me to limit to the right name. Change your system. Please contact me Your system is absolutely client unfriendly. I have to do a whole lot of things while I am only asking to change a mistake from your side. Tell me WHY I can not even submit this feedback Why do I have to sign in again and again?

user670f88ac64a64 1 month ago New Features

a month ago

0

efrainsolarte

Is there a reasonable way to define how my articles should be searched?
Randomly mixing up names is not appropriate and leads to confusion.
Is it not possible to remove from the list of mixed names those that are not appropriate or not used?
If there is such a procedure, please let me know.

23 days ago (edited)

0

reply

user673608614a7ed

I am wrongly cited as a co-author in the article - "DANIEL SANTANA SOARES TCC (Planejamento e Compras Publicas)". I was just cited in the reference.

7 days ago (edited)

0

reply

hannah_ehrlich

Merged in a post:

1

Another suggestion to fix the broken mentions feature.

Since I expect this is one of the main features that users would like to use, it is remarkable that academia.edu have not fixed it for so many years - especially when it's not much development work to do so. Another fix would to be allow people to say yes or no to suggested mentions en-masse or in bulk. This would be easy to do in various ways but one easy tried and tested solution is to put a check box next to each entry in the list of suggested mentions and then allow people to check all for say 100 at a time plus check/uncheck manually and then select yes or no for all checked entries.

jfrmasters 1 month ago Product Improvements

18 hours ago

0

Got ideas for new features?

Changelog

Feedback