arunsupe

Very interesting list of deceptive influences. Thanks for sharing.

Thanks for your comments.

I am finding myself getting sucked into the decisiveness and click baiting nowadays. Want to implement some of these counter measures. Here are my thoughts: 1. DDG or searxng for search YouTube only via the search engine (not YouTube's recommendations - worse than tiktok) 2. Reuters for news (no CNN, CNBC, MSNBC, fox, Google, Yahoo) 3. No social media (don't do this much anymore already) 4. Turn off personalization for Google (I use this the most)

This should limit a large percent of external influence. Any other thoughts/suggestions?

I understand your point about “boring”. But, I wonder if I will form my own interests if given more balanced options.

Hehe. You are right. There are many leftist media outlets.

But my point was not to be partisan; just to note that there will be lots of news to cover. Trump is a controversial figure. Every news outlet will profit. Now is not the time to downsize.

Not sure why. Criticizing the republican (especially Trump) government will be hugely profitable - just under 50% of the US population (175M people) is looking for such criticism, and there is no one else to do it. Every day there will be many opportunities to discuss/analyze/criticize/mock/fact-check/challenge.

They should be upsizing, not downsizing. My $0.02.

My question was tongue-in-cheek. I feel the Nobel committee, like every other organization, is also trying to stay relevant and topical. Hard to do this by giving out prizes to old scientists from obscure fields, no matter how important their findings. Hence, a congratulatory prize for the first black president of the USA, a consolation prize for another for losing to Bush, now prizes for computer scientists at the center of the AI fad... An influencer is the logical endpoint for this trend. Other examples of such slipping-down-the-slope: "Time's person of the year" is always a person in the limelight, Oscars favor blockbusters rather than movies with great acting, ditto with music awards, companies that rebranded to the .com fad or the iDevice fad...

:-)

Seems intuitive, but is there any evidence to support it?

Great post. It's nice that these rules can be trivially demonstrated by simulation. The simulation (and visuals) helps validate the concepts.

Added support for multiple languages, using fasttext's embeddings

Correct.

I assumed you meant computer languages :-) If you mean human languages, yes Google publishes word2vec embeddings in many different human languages. Not sure though how easy it is to download.

This implementation does not because the query has to be a word. One way to extend it to phrases is to average the vectors of each word in the phrase. Another way is to have a word2vec model that embeds phrases. (The large Google News model I think does have phrases, with spaces converted to underscore). But going from words to phrases opens up a whole new can of worms - how large a phrase to consider from the input stream etc. Plus, I don't think averaging words in the phrase is the same as learning the embedding for the phrase. Sentence embedding models are necessary for that, but they are far too slow for this use case as pointed out by others.

To summarize, this is a simple implementation that works for the simplest use case - semantic matching of words.

I would think translation to other languages would be trivial. The model is just a map of word to vector. Every word is converted to it's vector representation; then the query word is compared to the input words using cosine similarity.

The model in Google drive is the official model from Google and will work.

Haven't tried the huggingface model, but, looks very different. Unlikely to work.

Oh.Please explain. The example was entirely arbitrary. Should I change it?

I've built a command-line tool called semantic-grep for searching free-text English content using natural language queries. Some key features: Uses word2vec embeddings to find semantically similar text Written in Go, using only the standard library Familiar CLI UI Can process data streams piped as input Finds similarity in a sliding window of text It's designed for developers and researchers who need to search large text datasets more intelligently than simple keyword matching. The tool is still in early stages, so I'd appreciate any feedback or feature suggestions. GitHub: https://github.com/arunsupe/semantic-grep